1

Department of Chemical Engineering

The City College of New York

CHEMICAL ENGINEERING LABORATORY MANUAL

General Procedures and Practices

For ChE 360, 460, and 462

CCNY EMERGENCY TELEPHONE NUMBER: 7777

Prepared by Prof. C.A. Steiner. July, 2004

(revised January 2008)

ACKNOWLEDGMENT

On behalf of both the Department and myself, I wish to acknowledge the efforts
of both Chief lab Technician Zhen Rong Xu and Senior Lab Technician Andrew
Eng. Their contributions to the three Chemical Engineering laboratory courses
cannot be overstated. Moreover, their participation in the actual operation of the
lab and their interaction with our students contributes immeasurably to the quality
of the learning experience.

C. Steiner
January, 2008

TABLE OF CONTENTS
Part I. General Laboratory Practice
Introduction..1
Purpose of this Manual....1
Sample System: The Refrigerator Problem.....2
Chapter 1: LABORATORY SAFETY...3
Safety Rules..3
Before you start your experiment...3
While running the experiment.....5
When you are finished working for the day..7
Potential Hazards.7
Safety Audits.7
Why conduct a safety audit?..7
How is a safety audit conducted?..8
Checklist for safety audits...9
Chapter 2: PLANNING YOUR EXPERIMENT.13
General Outline...13
Designing your experiment...14
Setting up your experimental apparatus.16
Designing your procedure.17
What comes next?..19
Chapter 3: THE LABORATORY NOTEBOOK....20
Purpose of the laboratory notebook20
What goes into a laboratory notebook?..21
Chapter 4: RUNNING THE EXPERIMENT..25
Before you start..25
The experiment itself.25
Shutting down.26
Chapter 5: ANALYSIS OF DATA...27
Some definitions.27
What are data? What are results? What are conclusions?
And whats the difference?27
How good are your data? precision vs. accuracy and
reproducibility..28
How bad are your data? random vs. systematic error,
experimental error, and plain old mistakes.28
How to present your data significant figures and scientific
notation.30

Bases for comparison order of magnitude analysis,

sensitivity analysis, and efficiency...31
Statistical Analysis.32
Mean, variance, standard deviation, and confidence intervals
(student t-test).33
Linear regression analysis, standard error of estimate, and
correlation coefficients...33
Non-linear regression analysis, semi-log plots, log-log plots,
and the F test..34
Analysis of Data: Tables and Figures.35
Making Sense of Your Results.45
Chapter 6: THE LABORATORY REPORT...47
What are the different sections of a laboratory report, and what
goes in each section?47
Style, or how to write a good report.52
Sentence structure.53
Paragraph structure54
Some general hints for good technical writing...55
Bad words, or words to avoid or use with caution.56
Sample laboratory reports on the refrigerator problem56
Complete long version...57
Sample short versions...60
Chapter 7: UNIT OPERATIONS LABORATORY HARDWARE...63
Bibliography..64
General Reference Texts..64
Unit Operations Texts64
Statistics Texts64
Safety Texts.64
Texts on Good Writing...64
Suppliers and Catalogs.65

Introduction
Chemical engineering happens in the laboratory and chemical plant. In
the classroom courses in our curriculum you learn concepts and analytical skills
pertaining to the design of chemical processes and their interpretation in terms of
physical laws. In the laboratory courses you are called upon to put your
knowledge to work. Rather than labeling a schematic diagram of a piece of
equipment, you must label the equipment itself, which means you must learn
what each component looks like, how it works, and the importance of its position
(top vs. bottom, downstream from a pump vs. upstream, etc.) in the apparatus.
Rather than making assumptions about whether or not certain effects are
significant in your system, you must measure these effects directly to determine
their absolute magnitude, thereby putting your assumptions to the test. Rather
than deriving equations, you must apply these equations in the analysis and
interpretation of your results, and consider whether or not they provide a
complete picture of your experimental system.
There are very specific rules that apply when conducting laboratory work.
They cover topics ranging from safety in the handling of chemicals and
equipment to the correct use of statistics in designing experiments and analyzing
results. These rules and practices apply universally in the conduct of experiments
and research. Therefore it is essential for practicing engineers to master them
and always follow them as a matter of course.
Purpose of this manual
This manual is intended to be a general reference for all chemical
engineering laboratory and plant work. It contains chapters on laboratory safety,
planning and correct conduct of experiments, how to keep a proper laboratory
notebook, statistical analysis of data, preparation of figures and tables, and
writing laboratory reports. The discussions in these chapters apply to all
laboratory and plant work, and thus extend beyond the requirements of ChE 460.
In addition, there are chapters covering each experiment in ChE 460 Unit
Operations I Laboratory. These discussions include information about the basic
principles illustrated in each experiment, the equipment and its operation, and
study questions that are designed to help you analyze the experiment and the
equipment critically. In addition, students are expected to refer to Unit
Operations textbooks, sources such as Perrys Handbook of Chemical
Engineering, and catalogs from laboratory and industrial suppliers for more
background on each experiment and each component of the equipment. A
comprehensive bibliography is included.

SAMPLE SYSTEM: THE REFRIGERATOR PROBLEM

Purpose
To select a familiar system to serve as a model with which to illustrate
concepts presented in this manual. Examples based on this system will be
presented throughout the manual.
The Refrigerator Problem
Suppose you wish to calibrate the refrigerator in your home. Say the
refrigerator has 6 settings, numbered 1-6, with 1 as the warmest and 6 as the
coldest. Your primary goal is to determine the equilibrium temperature in the
refrigerator corresponding to each setting. As a secondary goal, you wish to
analyze the performance of the refrigerator with respect to the time it takes to
reach equilibrium when the setting is increased by 1 scale increment. In addition,
you suspect that the temperature within the refrigerator is not uniform, that is,
there appears to be a hot spot near the light bulb, and you have also noticed
that sometimes food in the back of the bottom shelf freezes. Therefore you wish
to analyze the effect of position on the temperature inside the refrigerator at
different settings.
Throughout this manual we will refer to the refrigerator problem to
illustrate concepts such as design of an experiment, taking measurements,
analysis of data, and writing a lab report. The refrigerator examples are indented
and in a smaller font than the rest of the text so they can be located easily.

Chapter 1: LABORATORY SAFETY

The importance of following strict safety procedures in the laboratory and
the chemical plant cannot be overstated. Every piece of equipment and every
chemical and gas may pose some kind of hazard under some set of
circumstances. You can minimize your risk and that of your coworkers by
educating yourself as to potential hazards, following strict safety procedures as
outlined below, applying common sense, and keeping your eyes and ears open.
This chapter is divided into three sections, headed Laboratory Safety
Rules, Potential Hazards, and Safety Audits of Laboratory Equipment. In
addition, the official CCNY Department of Chemical Engineering Safety Policy is
included at the end of the chapter. Laboratory safety rules include general rules
of conduct, proper work habits, and safety precautions that should always be
followed when working in a laboratory. Potential hazards include chemicals, high
pressure, high temperature, moving parts, sparks or other potential ignition
sources, breakage, etc. The section on hazards includes information on how to
minimize the risk from these factors. A safety audit of laboratory equipment is an
inspection of the equipment to be conducted prior to conducting each experiment
that is intended to familiarize you with the physical apparatus, its operation, and
its hazards. A Safety Audit Checklist is provided to insure that all possible
hazards are reviewed and minimized whenever you work in the lab.
SAFETY RULES
In order to ensure your safety and that of your co-workers, always follow
the safety rules given below. Note that these are rules, not suggestions!
Before you start your experiment:
1 Before you enter the laboratory, make sure you are properly dressed
and outfitted. Also inspect your coworkers clothes for hazards they might
have missed.
Put on safety goggles.
Remove dangling necklaces, ID cards, earrings, keychains,
bracelets, etc.
Check clothing for loops, baggy pockets, baggy sleeves,
dangling cords, scarves, ties, etc. and tie them back securely or
wear a lab coat to cover them.
Wear only closed-toe shoes in the lab (steel-toed safety shoes
if appropriate). Canvas shoes are not recommended and sandals
are forbidden.
Tie back long hair and long beards.
Turn off all cell phones and other personal electronic devices.

2 No food or drinks are permitted in the lab. Smoking is prohibited.

3 Turn on all the lights (learn the location of all the light switches).
Remember that natural light from the windows will fade at dusk!
4 Sign yourself into the lab. In case of an emergency, it is essential that
we know who is inside!
5a Take a safety walk through the laboratory.
Locate all exits. Make sure they are not blocked.
Locate all safety showers and eye washes. Make sure that
access to them is not blocked.
Locate all fire extinguishers. Inspect them to be sure they are
fully charged and properly secured, and that access to them is
not blocked.
Locate all fume hoods and make sure they are:
closed
working properly
Locate all continuous monitors for toxic gases that may be
installed in the laboratory and make sure they are working
properly and register safe levels of the gases they are designed
to detect.
Locate all telephones. Check to make sure they work. Know
the emergency phone numbers, and make sure they are posted
directly on the telephones. On the City College campus, the
emergency telephone number is 7777.
5b - Prepare a checklist of the items in 5a and their locations, and keep it
in a clipboard hanging by the door for the benefit of all others who use the
lab.
6 Never start working on an experiment until you have studied its
operation, including its start-up and shutdown procedures!
7 Assemble any required safety tools.
Protective Clothing
Goggles
Lab coat
Gloves
Heat resistant
Chemical resistant
Hard hat
Dust mask
Steel-toed shoes
Dark glasses for looking at bright lights
Ear plugs or earmuffs, as needed
Forceps for handling hot vessels, materials, etc.
Other (as required)
8 Inspect your work area.
Remove clutter from all surfaces.
Remove clutter from the floor.
Clean the bench top and the general work area.

Make sure that all components are secured in place and there
are no loose items in the area. Remove or secure any potential
spilling hazards, projectiles, tripping hazards, etc.
Make sure all switches, controllers, plugs, dials, gauges,
meters, etc. are readily accessible and easy to see or read. If a
unit is mounted too high for you to see or reach, get a stepstool
(not a chair!) before you start working!
Make sure all electrical wires are plugged in all the way.
Make sure you have all the required small (removable) devices
within reach, such as
Stopwatch
Ruler
Collection vessels (beakers, flasks, etc.)
Measuring vessels (graduated cylinders, etc.)
Thermometer
Magnifying glass
Stethoscope
Other
Make sure you have your lab book and several working pens
and pencils within reach. Make sure there is a clear area on
which to put your open lab book.
9 Familiarize yourself with the potential hazards, safe handling practices,
safe disposal practices, and emergency procedures for dealing with
accidental spills, leaks, etc. for every chemical you will be using! This
information is given in a document called a Material Safety Data Sheet
(MSDS), which by law must accompany every chemical shipped in the
United States. Every laboratory should have a file with an MSDS for every
chemical used or stored in that lab. Never use any chemical until you
have read and understood the warnings and recommendations in the
MSDS, and have arranged for whatever protective measures are
specified.
10 Familiarize yourself with all emergency shut-down procedures for
your equipment. This includes the sequence for closing valves, shutting
down power, etc. These procedures should be posted on the equipment,
described in your lab manual, or both.
11 Prior to running an experiment, always conduct a complete safety
audit on your equipment. A complete safety audit checklist is given at the
end of this chapter.

10

While running the experiment:

1 Use the buddy system never work in the lab alone.
2 Do not make any modifications to the equipment without the approval
of the instructor or technician responsible for the lab.
3 Do not make any modifications to the procedure without the approval
of the instructor or technician responsible for the lab.
4 Always be attuned to unexpected
Sounds
Hiss is there a gas leak?
Drip is there a liquid leak?
Rattle
is something loose?
Is something rotating off balance?
Gurgle
Is something bubbling over?
Is there an air pocket in a liquid line?
Hum is something stuck?
Sights
Does everything look right?
Bulges in flexible hoses or tubing might blow out!
(we had this happen one year! There was a big loud
BANG! The solution place a pressure control
valve in the line and use rubber hose thats properly
rated for that pressure level).
Does anything appear to be coming loose?
Is anything in danger of overflowing?
Other
Check all sensors at frequent intervals to be sure the
system is operating within acceptable ranges of
Temperature
Pressure
Flow rates
Malfunctions
Is any part of the system overheating?
Any unexpected vibrations?
Are all lines clear?
Check for kinks in flexible tubing.
Other
5 Inform the instructor or technician immediately if you have any
concerns.

11

6 If you take a break while working, do not leave the equipment running
unattended. Make sure it is being watched by someone who is familiar
with its hazards and its correct operation.
7 Clearly label all flasks, jars, etc. The label should include the
contents of the vessel, the date, and your name.
8 All toxic or volatile chemicals are to be opened and used only in
working fume hoods.
9 All chemical wastes must be disposed of properly. Do not pour any
unknowns down sinks or drains.
10 DO NOT ATTEMPT TO FIGHT A FIRE! In the event of a fire,
evacuate the lab and call 6911 immediately!
When you are finished working for the day:
1 - Shut down the equipment, following written procedure, if available.
Make sure the system returns to its initial state (dry, room temperature,
etc.).
2 - Tidy up the work area.
3 - Sign yourself out of the lab.
POTENTIAL HAZARDS
There are many potential hazards in any chemical laboratory or chemical
plant. Some of these can often be removed without compromising the process or
products. In some cases it may be possible to redesign a system so as to
eliminate certain hazards. Other times these hazardous materials, system
components, or practices are a necessary part of the operation of the system. It
is essential that you learn how to identify these hazards and how to minimize the
risk that they pose. Simple measures such as insulating hot equipment or using
the proper containers for corrosive chemicals can significantly reduce the risk of
injury or accidents. Some common hazards found in chemical labs and plants
are presented in the section on Safety Audits, below. Appropriate measures for
their handling include the use of properly rated equipment, vessels, tubing,
protective clothing, etc., and should be reviewed and implemented each time a
potential hazard is encountered.

12

SAFETY AUDITS
In this course you will be required to prepare a written safety audit for
each experiment. This report is to be handed in at least 2 days before you
conduct the experiment. No student will be allowed to work on a piece of
equipment unless he/she has handed in the appropriate safety audit.
Why conduct a safety audit?
Even though most labs and industrial settings have strict procedures in
place regarding the start-up, use, and shut down of every piece of equipment, it
is still the responsibility of each user to review the condition of the equipment
before starting operation. Remember that all users are human. If the last person
to use the equipment overlooked some potential hazard, you may be the one to
suffer an injury. Here are 2 rules to live by:
1 NEVER PUT YOURSELF AT RISK!
2 NEVER PUT ANYONE ELSE AT RISK!
How is a safety audit conducted?
On the following pages there is a checklist that may be used for your
safety audits. The checklist covers potential electrical, fire, pressure, chemical,
mechanical, thermal, breakage, and other miscellaneous hazards. You should
also add other types of hazards if you encounter them. If you do, please inform
the instructor so that we may update this checklist. In addition, your safety audit
must include a hand-drawn schematic of the equipment. Make sure this diagram
shows the sources of all inputs (building utilities, boiler in the lab, etc.), giving
their pressure, temperature, and flow rate, as appropriate, and also the
destination of all outlet streams (city sewers, city air, special receptacle, etc.).

13

CHECKLIST FOR SAFETY AUDITS

General equipment inspection
Are all units and devices clearly labeled?
Are all controls and gauges free of obstructions and clearly visible and
accessible?
Are start-up and shut-down procedures posted in the work area?

Types of hazards

Electrical
Insufficient or frayed insulation
Incorrect fuses
Broken plugs
Any exposed wiring
Equipment not grounded
Fire
Sources of sparks?
Electrical switches
Electrical heating elements
Pilot lights
Static electricity
Ungrounded appliances
Sources of fuel?
Uncontained vapor
Flammable liquids
Loose paper or other combustibles
Sources of oxygen?
Open systems the air!
Closed systems air leaks!
Chemical
Chemical safety MSDS (see section on Potential Hazards,
above)
Skin?
Eyes?
Inhalation?
Ingestion?
Chemical interaction with
Clothing?
Containers?
Other materials in the vicinity?
Plastics
Bench top
Insulation
Other

14

Pressure
What gases are present?
Compressed air
Steam
Nitrogen or other inert carrier gases
Vapors
Other
What is their source?
Building utility
Boiler (location?)
Cylinder
Vessel
How is the gas pressure regulated at the source?
Some remote pump (basement?)
Temperature controller on the boiler
Regulator on cylinder (is it the right kind?)
Vessel cap (what is the rating of the vessel and the cap?)
Where does the gas flow? Check pressure ratings on
Tubing
Regulator(s)
Equipment
Vessels
All valves
Is the gass pathway unobstructed?
Is the gas properly contained and/or isolated from sources of
ignition or chemical reaction? (MSDS!)
Check all valves, seals, joints for leaks
Mechanical
What moves?
Is it bolted down?
Is it enclosed/encased?
What safety precautions should people take?
Clothing
Hair
Jewelry
Thermal
Whats hot?
How hot?
How controlled?
Whats cold?
How cold?
Any freezing hazards? (remember that plasticware and
rubber hoses all undergo a glass transition at some
temperature characteristic of the specific material).
Breakage
Glass components check for cracks

15

All glass pressure and vacuum vessels should be taped

Rubber hoses (look for signs of wear or disintegration)
Miscellaneous
General housekeeping
Any tripping hazards on the floor?
Clutter on the bench top?
Are all parts of the equipment secured in place?
Is the work area well lit?
Is there a clear path to the nearest exit?

Proper disposal of spent materials

Chemicals
What are they? (Note the contents of all containers, flasks,
jars, etc. should be clearly marked on the vessel).
What is the proper disposal procedure (reference?)
Are all the elements of this procedure available and in place?
Describe/locate.
Proper labeled receptacles
Proper waste cans
Vapors
What are they?
Are they contained in appropriate leak-proof vessels? Verify,
stating what material the vessels are made of and citing
references indicating that they are suitable.
What is the proper disposal procedure (reference?)
Are all the elements of this procedure available and in place?
Describe/locate
Proper labeled receptacles
Proper waste cans
Biohazards
What are they?
What is the proper disposal procedure (reference?)
Are all the elements of this procedure available and in place?
Describe/locate
Proper labeled receptacles
Proper waste cans
Sharps (razor blades, hypodermic needles, knife blades, etc.)
What are they?
What is the proper disposal procedure (reference?)
Are all the elements of this procedure available and in place?
Describe/locate
Proper labeled receptacles
Proper waste cans

16

Disposable glassware
What are they?
Pasteur pipettes
Capillary tubes/pipettes
What is the proper disposal procedure (reference?)
Are all the elements of this procedure available and in place?
Describe/locate
Proper labeled receptacles
Proper waste cans
Disposable plasticware
What are they?
Transfer pipettes
Hypodermic syringes
Other
What is the proper disposal procedure (reference?) Note this
will depend on what chemicals or biological agents were in
contact with the plasticware
Are all the elements of this procedure available and in place?
Describe/locate
Proper labeled receptacles
Proper waste cans

17

Chapter 2: PLANNING YOUR EXPERIMENT

General Outline
What is the general rationale for this experiment (applications, questions
you seek to address, etc.)?
What is your objective (be specific)?
What are the
Inputs?
Outputs?
Settings (parameters)?
What is the range of values accessible or of interest for each of the items
above? Explain.
How many measurements do you need to make
How many different settings?
How many repeat measurements at each setting?
Exactly what settings will you use, and why (what question(s) will this
specific set of experiments address?)? Try to come up with a familiar system
characterized by the same quantities, so you have a physical feel for your
system.
How much time does each run take? Will you be able to complete all your
runs in the time allotted, and still have time to clean up after?
Equipment
Is it appropriate? Can you measure everything you need to?
Is it in good working order?
Do you need any additional devices, such as stopwatch, thermometer,
collection vessel, etc?
Did it pass your safety audit?

When planning an experiment, you must consider what your objectives

are, what equipment you have available, and what the limitations are of the
equipment. You must also consider what parameters (i.e. fixed experimental
conditions) are likely to influence your results. The more factors you decide to
analyze, the more complex your experimental plan must become, in order to take
all those factors into consideration. Next, you must design appropriate control
experiments. Control experiments provide a baseline for your system, giving
quantitative information about the response of the system independent of the
experimental factors you are studying. Finally, you must consider what the best
way is to measure your response. This includes selecting appropriate equipment
and measuring devices, using them in an appropriate and well-calibrated
manner, possibly designing two or more different types of experiments to provide
independent verification of your results, and also making sure that the output
from your instrument is what you think it is. This may sound strange, but think for
a minute about how most instruments work. In general, the instrument senses
something about the experimental system, converts that response to an electrical

18

signal, then uses the electrical signal in a software package that performs some
calculation and gives a result. Even though the computer doesnt know what that
result means, it is programmed to call that result Property X, for example. Be
careful! The computer may tell you, for instance, that Property X = -2.0. If you
know that Property X must have a positive value, then dont think youre finished!
You have to delve more deeply into what the result of the computers calculation
actually means! Maybe the value of Property X for your sample is actually
outside the range of your instrument. Maybe your sample isnt characterized by
a Property X at all. For example, a solid is not characterized by a viscosity. Yet
if you place a solid in a viscometer, e.g., and subject it to shear, it will generate
some signal that the software will then use to perform the calculation it is
programmed to perform, and display the result under the heading viscosity,
because thats what the program does. Thus it is important to make sure that
you use all instruments correctly, with appropriate types of samples. And if your
results look suspicious, proceed with caution. Perhaps a different type of
instrument should be used.
Here is an example that illustrates how a seemingly simple objective can
require a very extensive experimental plan. Suppose you wish to determine the
effect of solute concentration on the viscosity of a solution of solute A in water.
Clearly, you will have to make up a series of solutions, each with a different
concentration of A in water, and measure the viscosity. Now ask yourself what
experimental conditions (parameters) might play a role in your results. Perhaps
temperature will thus you should either select a single temperature at which to
make all your measurements (making sure that your temperature controller is
adequate to maintain your solutions at that temperature), or measure the
viscosity of each solution at several different temperatures to see if temperature
has an effect on your results and to calibrate the magnitude of that effect. Is it
possible that heat has an irreversible effect (hysteresis) on the viscosity of the
solution? Then its probably a good idea to make repeat measurements on the
same solution, first at successively higher temperatures and then after cooling, to
calibrate this effect. What is the time required for the viscosity to reach its
limiting (equilibrium) value after heating? Perhaps your experimental design
should include elapsed time at each set temperature as a variable. If the solution
is non-Newtonian, you should probably measure the viscosity at different shear
rates. If the solution is very viscous, you may want to measure its viscosity once
a day for a few days after mixing to make sure it is at equilibrium. You can see
how a seemingly simple problem can become quite complex!
Designing your experiment
The best approach to designing an experiment is to try to answer the most
questions with the fewest measurements. Start by asking the following
questions:
1 - What property or properties (y) do I want to measure? What is the expected
range of these values? The properties you measure are the output variables, y
(also called dependent variables). Note the y in boldface type indicates that y is

19

a vector of the form y1, y2, y3, that is, there may be more than one output
variable and each one must be considered separately in designing the
experiment and analyzing the results. Make sure all your equipment and
instruments are appropriate for the range of y you will be working with.
2 What is the independent variable, x, in my system? Over what range of x do
I want to make my measurements? The independent (or input) variable, x, is a
property of the system that is set by you. Typically you will make measurements
of y at different values of x. Choose a range of x for your study that addresses
your needs. Theres no point in measuring, say, the effect solute concentration
(x) on solution viscosity (y) at concentrations above the limit of what youre ever
going to use. Note - in some cases the independent variable is time (t).
3 What experimental conditions (parameters z) might be important in my
system? Over what range should I set these parameters? Experimental
parameters such as temperature, pressure, sample handling variables (such as
stirring time, stirring rate, etc.), etc. may contribute to the response of your
system, hence the value of your output(s) y. Select those parameters that are
important for your purposes, and select a reasonable range of parameter values
to study based on your present and future needs.
Once you have decided on the range to use for each of the inputs x and z,
you must select intermediate values of x and z within these ranges at which to
make your measurements, with the goal of obtaining enough data points to be
able to identify trends in each output y as a function of x and z. Typically 3
values for each parameter z and 4 or more (depending on how complex the
function y(x) is) for the independent variable x make for a good start. The formal
procedure for designing a series of experiments in which more than one factor, or
parameter, is studied simultaneously is called factorial design of experiments,
and is covered in detail in statistics texts (see Bibliography). After completing
this first set of experiments, inspection of the data will tell you whether more
measurements at additional values of x and/or z are warranted (see Analysis of
Data, chapter 5).
4 What control experiments should I run? A control experiment is one in which
you analyze the performance of a sample or a system in the absence of the
effect you are investigating, with all other conditions being the same. For
example, suppose you want to determine the influence of solute concentration on
the viscosity of solutions of solute A in water at 3 different temperatures, T1, T2,
and T3. To design appropriate control experiments, consider the system without
the experimental variable, that is, take away the solute! Measure the viscosity of
water alone at each of the 3 temperatures of interest.

20

Setting up your experimental apparatus

Your experimental apparatus may be designed and built by you or you
may make use of an existing apparatus with appropriate modifications, as
needed. Ask yourself the following questions:
1 What is the best method for measuring my outputs, y? There are often
different ways to measure the same property of a system, with different
advantages, disadvantages, and limitations to each. Before setting up an
experiment, familiarize yourself with as many of these options as possible,
making sure you learn what each method senses, how this information is used,
what is the useful range of each method, etc. Then select the best method for
your system. Some examples of different types of measuring devices are
covered in Chapter 7 Unit Operations Laboratory Hardware, as well as in
references such as Perrys Handbook and also catalogues from lab equipment
and hardware manufacturers and distributors (see Bibliography).
2 What other equipment, vessels, etc. do I need? Draw a schematic diagram
of your experiment and identify your needs with respect to vessels such as tanks,
columns, etc.; sample handling equipment such as stirrers, agitators, etc.;
components in a flowing stream such as pumps, valves, stream splitters (Ts, ys,
etcl) tubing, pipes, hoses, clamps and fittings; heat transfer equipment such as
heaters, heat exchangers, condensers, and evaporators; etc. For each
component, determine:
a. What is the appropriate rating (with respect to temperature limits,
pressure limits, capacity, etc.) for every component in my system?
All laboratory vessels, hardware, hoses, tubing, fittings, and equipment
are rated with respect to the maximum (and, if appropriate, minimum)
temperature, pressure, and capacity at which they can be used safely.
Consult catalogues (see Bibliography) and manufacturers literature for
information on the specific components in your system. Identify the
weakest link in your apparatus (is there a valve thats rated for a lower
pressure than your tubing and vessels? Which fitting or component has
the lowest temperature rating? Even if all your hoses and tubing are rated
appropriately for your needs, are they connected with appropriately-rated
fittings?) Be sure you can operate your apparatus safely under the
conditions of your experiment.
b. If I am using an existing apparatus, are all the components, fittings, etc.
rated for my needs? If not, which ones must be replaced before I can
conduct my experiment? Also make sure you check the existing
apparatus for wear, corrosion, or any other signs of aging.

21

c. What are appropriate (and inappropriate) materials of construction for

each component? The materials of construction dictate not only the
temperature limits for safe operation of a component, but also the
corrosivity of a component in contact with other chemicals, gases, vapors,
or water. Make a list of all chemicals used in your experiment (including
the products of any chemical reactions that might occur) and check to
make sure that they are safe for use with every component in your
system. Here again, refer to manufacturers literature or supplier
catalogues for the required information.
3 What safety checks should I perform before operating at full capacity?
Always make sure to leak test all connections, check all controllers (temperature,
pressure, flow, etc.), inspect all insulation, etc. before running an experiment (see
Laboratory Safety, chapter 1).
Designing your procedure
Outline a step-by-step procedure for your experiment. Think through the
details of how to proceed, such as whether to vary x from lowest to highest or
vice versa, what time intervals you should use, what the criterion is for your
endpoint, which steps to repeat, etc. Write out a detailed procedure directly in
your laboratory notebook. This procedure should be followed each time you
repeat the experiment. If you make any modifications to the procedure, they
should be noted and dated in your lab book as well. The first step or steps
should describe how you set up your experimental apparatus. Include all
relevant details. The second step tells how to initialize your system. This
includes the required initial settings for all indicators and valves, the required
initial conditions for the system (temperature, pressure, flow rate, etc.), etc.
Steps 3 and up contain a step-by-step procedure for running the experiment (see
the refrigerator example, below). The steps should explicitly state what
information is to be recorded at each step, such as time, ambient temperature,
parameter values, readings, etc. Note that the procedure can take the form of a
flow sheet, where a setting is changed after completing a run and then the
previous steps, identified by number, are repeated. Writing out a procedure in
this way really focuses your thinking about how to run the experiment, what
details you must be sensitive to, and how to instruct someone else on conducting
this work. Preparing written instructions of this form is essential for professional
engineers, who generally must supervise and direct one or more technicians who
actually run the experiment according to the engineers instructions.
Lets look at the refrigerator problem. The objectives are to
correlate the equilibrium temperature (output y1) and time to equilibrium
(output y2) with setting (independent variable x) and position in the
refrigerator (parameter z). Should the refrigerator be full or empty during
your experiment? Empty provides the best control over your conditions,

22

since any contents will give off heat and also interfere with the circulation
of air inside the refrigerator. Based on your observations of the
performance of the refrigerator, you selected 3 different positions in the
refrigerator at which to take your measurements, viz., position 1 top shelf
next to the light bulb (suspected hot spot), position 2 middle shelf near
the front, and 3 bottom shelf near the back (suspected cold spot),
hence you will need 3 thermometers, one at each position. According to
the manufacturers literature on the thermometers you are using,
temperature measurements of liquids are more reliable than of air, due to
the possibility of condensate forming on the thermometer bulb and
removing heat from the surface. Therefore you decide to place a glass of
water at each position and keep the thermometer bulbs immersed in water
at all times. However, remember that water has a high heat capacity (that
is, a high thermal inertia), so there will be some time lag between when the
air in the refrigerator reaches its equilibrium temperature and when the
water does. Moreover, the time lag will depend on how much water is in
the glass. Therefore, in order to be able to make a valid comparison of the
rates of equilibration at the three positions, you will have to make sure that
you have the same volume of water, in identical glasses (with the same
shape and same heat transfer properties), at each position. In order to
minimize the time lag between the air and water equilibration times at each
setting, keep the volume of water to a minimum by using narrow water
glasses, such that even when the water is deep enough to cover the
thermometer bulbs, the total volume of water is still relatively low. It is also
important to keep the glasses covered at all times, to prevent evaporation.
Thus you will have to use (and possibly also design) covers for the glasses
that are vapor-tight but that have a hole into which a thermometer may be
inserted without a gap, such that the numbers on the thermometer are
R
clearly visible without opening or uncovering the glass of water. Parafilm
or some other commercial plastic wrap, molded around the thermometer
and stretched and sealed over the top of the glass, could probably serve
this purpose. To ensure that all 3 glasses have identical heat transfer
properties, they should all be made of the same material, with the same
thickness, and preferably all from the same lot. The next thing you have to
decide is whether to vary the dial setting (your independent variable x)
from lowest to highest or highest to lowest. When you turn the dial up 1
setting, you turn on the compressor and make the refrigerator work
actively to cool its contents. On the other hand, when you turn the dial
down 1 setting, the refrigerator stops working and its contents become
warmer passively. Thus since we are interested in the refrigerators
performance, lets analyze it in an active mode by studying the equilibrium
temperature as we increase the dial setting. The next variable is time.
Each time we reset the dial, we will measure the temperature at all 3
positions at time intervals of, say, hour. When 3 repeat measurements,
hour apart, give the same result, then we will say that the system has
reached equilibrium. Note that if the first reading we take, hour after the
setting was changed, turns out to give the equilibrium temperature, we
have not succeeded in determining the time to equilibrium (y2). We must
then either repeat the experiments, starting from the same initial conditions
and making measurements at shorter time intervals, or decide that it is
sufficient for our purposes to know that the time to equilibrium is less than
or equal to hour. One more question how should we measure the
time? Either we must record the actual time, read off a clock or wristwatch,
and then calculate the elapsed time and record that as well for each data
point, or we must use a stopwatch and start it going at time t0 for each
setting and record the elapsed time with each reading. Lets choose the

23

stopwatch (make sure you have one before you start!). Now for the stepby-step procedure. Step 1 Empty the refrigerator completely. Then place
identical glasses of water, all filled to the same level and covered to
prevent evaporation of water, at each of the 3 test positions in the
refrigerator with a thermometer in each glass such that the thermometer
bulb is completely immersed and the scale on the thermometer is clearly
visible. Step 2 - set the initial condition by turning off the refrigerator and
allowing its temperature to reach ambient temperature. Record the
ambient temperature. Step 3 set the dial to setting 1 and close the
refrigerator door. Turn on the stopwatch and record the elapsed time as 0.
Step 4 Wait hour, open the door, and read all 3 thermometers as
quickly as possible so as not to leave the door open too long. Record all 3
temperatures, along with the elapsed time, in a data table as a function of
position. Repeat this step until all 3 temperatures have reached a constant
value on 3 successive readings, hour apart, making sure not to open the
refrigerator door between readings. Step 5 Increase the scale setting by
1, reset the stopwatch to 0, and go back to Step 4. Repeat for all scale
settings.
Are we done designing the experiment? Lets check weve
identified our ys (equilibrium temperature and time to equilibrium), xs
(scale setting and time), and z (position in the refrigerator). Weve decided
which way vary x (i.e., from lowest to highest), decided on our intervals in t
(at least to start, we may change this as we learn more about the response
of the system), decided what measuring instruments (thermometers and
stopwatch) to use and the best way to use them, and selected vessels with
the appropriate geometry and heat transfer properties. And weve written
a complete step-by-step procedure. Good job weve covered all the
points in the checklist above!

What comes next?

Now that you have designed your experiment, you must carry it out. Inspect
your data as you go along to make sure that your experiments actually address
the questions you sought to answer; if not, you may choose to modify your
procedure. For example, maybe your time intervals are too long (or too short) to
provide sufficiently detailed information. Maybe the range of inputs, x, you
selected is too narrow to show a statistically significant variation in y. You may
find it is a good idea to change your conditions or make additional measurements
in order to accomplish your goals.

24

Chapter 3: THE LABORATORY NOTEBOOK

The laboratory notebook should contain the complete primary record of all
information pertaining to your experiments. This includes not only the details of
what you did, but also all of your thoughts related to your experimental system,
what motivated you to conduct your investigation, your ideas about how to set up
and execute your experiments, drawings of your apparatus, your step by step
procedure, all data, plots of data, and your thoughts on the meaning of your
results. Remember that the notebook is a legal document that serves as proof of
what you actually did, and when.
Purpose of the laboratory notebook
The laboratory notebook is recognized as the primary record of everything
you planned, set up, did, measured, observed, and thought about pertaining to
your experiments. Primary record means that the information was recorded
directly into the lab book, not written down elsewhere first and then copied over.
In addition, every page should be dated and signed. There are several practical
reasons for all this. First, it puts all your thoughts together in one place, so no
potentially important information can be lost or forgotten and everything is
together when it comes time to analyze your results or review what you did.
Second, the lab book is a legal document. If there is ever a dispute about the
ownership or precedence of any work you say you did, the records in the lab
book establish exactly what you did, and on what date. This is extremely
important in industry and research, where new discoveries may be patentable.
Moreover, a lab book is not kept properly, it may not be admissible as legal
evidence to support your claims of ownership of the information recorded therein.
In industry, laboratory notebooks also provide a basis for the internal
designation of all samples. Lab notebooks are generally numbered with an
internal code, and all the pages are numbered as well. Thus, for example, the 3rd
sample recorded on page 12 of notebook 4864B has the in-house designation
4864B-12-3. Anyone using that sample can look up how and when it was made,
etc.
Laboratory notebooks are also important historical records. The
notebooks of Leonardo da Vinci, Marie Curie, and Thomas Edison, among
others, are still the subject of much study by historians, who are interested not
only in learning about the specifics of what was recorded in the notebooks, but
also in following the thought processes of these brilliant minds.

25

What goes into a laboratory notebook?

1 page numbers. Start by numbering all the pages, front and back. Leave the
first 4-6 pages blank they will ultimately be used for your Table of Contents.
Note that it is best to use a notebook with graph paper pages.
2 date write the date at the top of every page. Dont include work from more
than one date on any given page.
3 - room temperature Keep a log of the room temperature underneath the date
or alongside the data. Record the temperature and the time it was measured
several times over the course of the experiment, especially if the temperature
seems to change over the course of the day.
4 - Title each page should have a title. The title should state what experiments
are recorded on that page, what apparatus is drawn, etc.
5 Objective, hypothesis, and background related to the work on that page. For
example, for the objective, Investigate the relationship between solute
concentration and solution viscosity in solutions of Solute X in water as a function
of temperature, the hypothesis may be, Hypothesis: the viscosity will increase
with increasing concentration, and decrease with increasing temperature.
Include citations from published work that you used to formulate your hypothesis.
Also include predicted values, if available. For example, if you are measuring the
equilibrium vapor pressure of a liquid as a function of temperature, include a plot
of the expected values calculated using the Antoine equation constants for that
liquid. You can then compare your own data to this plot to determine whether
your system is indeed at equilibrium under your experimental conditions, or
whether you need to revise your experimental procedure to obtain the desired
conditions.
6 Experimental plan State briefly what the experiment will consist of, and be
specific about the values of the experimental parameters and how the data will
be used to meet the objectives. For example, for the objective and hypothesis
given above, the Experimental Plan might read, Prepare a series of 5 solutions
of X in distilled water, at concentrations of 0.5, 1.0, 1.5, 2.0, and 5.0% X by
weight, and measure their viscosity at T=20oC, T=30oC, and T=40oC. Determine
whether viscosity is sensitive to solute concentration under these conditions.
7 - Description and diagram of apparatus, including all manufacturers
information (make, model number, manufacturers name, address, and contact
information) for all commercial instruments used.

26

8 Materials
- for commercial chemicals, include all manufacturers, catalog numbers
and lot numbers
- for proprietary chemicals (that is, chemicals that are made by your
company but are not yet commercial products), include the in-house
designations, as appropriate
9 - Step-by-step procedure. The procedure in the lab book should serve as a set
of instructions for someone trying to replicate exactly what you did. It should
have the following features:
a. Detailed step by step instructions, with numbered steps and a decision
tree, if appropriate. Details should include information such as stirring
time, standing time, the settings used on each instrument, how you will
decide when a step is complete (e.g., The system is considered to be well
mixed if there is no turbidity visible to the naked eye when the vessel is
held up to the light.), etc., as appropriate. For example, the sample
preparation procedure for the work described in the Experimental Plan,
above, could read:
Sample preparation:
1 Weigh out 0.5g of X on the ABC Analytical Balance.
2 Add the measured quantity of X to 100 mL distilled water in a
250 mL Pyrex beaker.
3 Stir on the Brooklyn Model 1118 Magnetic Stirrer for 10 minutes
on setting 5, using a inch TeflonR stirring bar.
4 If the solution is cloudy, continue stirring for another 10 minutes.
If not, proceed to the next step.
5 Remove the solution from the stirrer and cover.
6 Repeat steps 2-5 for the remaining samples described in the
table below, using the weights given in Column (c) in 100 mL of
water.
Then prepare a table giving sample designation (column (a)),
concentration (column (b)), and required weight of Solute X for 100 mL of
water (column (c)).
b. Redundant instructions. As a back-up in case something is written
down incorrectly or is hard to read, each step should give the same
information in two different ways whenever possible, stating what to do
and what its purpose is. For example, Open valve 2 to allow air to flow
into the system from the compressor, is preferable to Open valve 2, or
Let air flow in from the compressor because it allows the reader to detect
any mistakes in advance (suppose the compressor valve is actually valve
3, not valve 2!) and to correct them before proceeding.

27

c. Refer to the diagram of the apparatus, for the reasons just noted.
10 Tables providing details of the experimental samples or runs. Typically you
will conduct experiments on more than one sample or run on a given day. Set up
a table to identify them. The first column should be headed sample number or
run number, followed by columns that give information about the conditions
(temperature, pressure, etc.), instruments settings, composition, etc.
distinguishing the entries. This table should be cross-referenced in the step by
step procedure, as noted above.
11 Raw data and data tables. Every number that you measure or read off an
indicator should be written in the notebook exactly as you see it in your
apparatus. You should also leave space in the notebook for values you calculate
using these measured values, but the actual measurements should always be
recorded directly into your notebook. This is the best way to avoid errors in your
final results!
In general, data should be recorded in a data table. The first column of
every data table should give the run number or sample number. The last column
of every data table should be for Comments. While you may not write an entry in
the Comments column for every experiment, this ensures, first, that you will
remind yourself to examine the experiment or sample for any unusual
observations or occurrences (e.g., there was a brief power surge during this
run, or the solution was cloudy, or note that this sample was stirred longer
than the others), and second, that you will write these observations down.
There should be one column for each input or parameter (temperature, flow rate,
concentration, etc.) and one column for each value that is measured during the
course of the experiment, and the column headings should all indicate the
appropriate units. In addition, you may wish to include a column or columns for
values that are calculated using these measured values. The headings for these
columns should indicate what value is calculated, and how. Also, every data
table should include a Comments column. This is where you record your visual
observations of your system and your environment (for example, did gas bubbles
appear here? Did the room temperature suddenly drop? Were there air currents
that affected your measurement? etc.).
Example. Suppose you wish to determine the pressure drop of air across a
packed column as a function of the flow rate of air through the system. Your
input is air flow rate, measurements are pressure upstream and downstream of
the column, and calculated result is the pressure drop, given by Pupstream
Pdownstream. The data table should be set up as follows:

28

(a)
run number

(b)
air flow rate
3
(ft /min)

(c)
pressure
upstream of
column (mm
Hg)

(d)
pressure
downstream
of column
(mm Hg)

(e)=(c)-(d)
pressure
drop across
the column
(mm Hg)

(f)
Comments

1
2
3
4
5
Note the use of lower case letters in parentheses to identify each column,
and the use of these letters in defining what is calculated in column (e). Using
this format it is easy to see at a glance which columns contain primary
measurements and which contain calculated results. An alternative format is to
include a footnote to the results column giving the equation used to calculate the
result.
12 All calculations. All equations and calculations should be written directly
into the lab book and annotated appropriately.
13 Plots of the data. Before leaving the lab for the day, prepare plots of the
tabulated data and results in the lab notebook. If possible, these plots should be
drawn directly under the data tables or on the facing page, so that if there are
any anomalous points or points that raise a question you can easily double check
them against the tabulated values, and also see if any unusual comments pertain
to the corresponding experiments. Also include a curve indicating predicted
values on the same plot (see item 5, above). If you see points that do not fit with
the general or expected trends, you may choose to repeat those experiments. A
detailed discussion on the preparation of plots is found in Chapter 5 Analysis of
Data.
14 General summary for the day. Compare the behavior shown on the plots
with what you expected based on your hypothesis. What did these experiments
show you? What additional experiments are suggested by these results?
15 - Dated signature on every page. This documents that you did this work on
this date. This information may be used to establish your rights to the work, if
you are ever challenged.

29

Chapter 4: RUNNING THE EXPERIMENT

Before you start
Check all components of equipment
Be prepared! Make sure you have all required portable devices on
hand, such as stopwatches, collection vessels, magnifying glasses,
stethoscopes, gloves, forceps, etc.
Start up equipment and bring to desired operating conditions
Temperature
Pressure
Flow rate
Other
Monitor each of these conditions for fluctuations over time scales of
order 5 minutes or so, and note the frequency and amplitude of these
fluctuations in your lab book. This information may be required for a
sensitivity analysis (see Ch. 5 Analysis of Data).
Check all joints, seals, etc. for leaks
Repair or replace as needed before continuing
The experiment itself
Enter all information (run number, settings, inputs, readings, etc.) in the
data table in your notebook
Enter all comments in your data table. Comments may be related to:
the experimental system itself
appearance (cloudy, bubbly, precipitated, etc.)
behavior (flooded, evaporated, etc.)
properties (temperature changes, pressure changes,
etc.)
timing (left it on all night, didnt allow system to come to
equilibrium, used different intervals between
measurements)
etc.
the conditions of the room
changes in temperature
power surges
etc.
observations regarding personnel or technique
was the run performed by a different person?
was there any difference in the technique used?
etc.
Inspect each data point for anomalous behavior.
Prepare plots as you go along.
Repeat any experiments that give anomalous results. Repeat yet again
for consistency.

30

Note down any modifications to your procedure in your lab book as you
work. Provide details of any decisions and/or any tricky parts to your
procedure that were not obvious to you before you started. Have
someone read this who has not done the experiment to see if they can
understand what you meant.

Shutting down
Make sure to follow the appropriate shutdown procedures, in order.
Check that the system has returned to its resting temperature, etc.
Clean up work area.
Sign yourself out of the lab.

31

Chapter 5: ANALYSIS OF DATA

OK, so you planned your experiment, you managed to carry it out
(hopefully without blowing up the lab!), you entered all your measurements in
your lab book, and now youre just about done, right? Wrong! Now comes the
real work. This is when you apply your intelligence, training, and insight to
realize the full value of the effort you have put into your experiments. And it must
be done carefully. Unless you analyze your data correctly (and there are strict
rules governing this process) you might end up drawing false conclusions from
your experiments.
In this chapter you will learn how to handle data properly. This includes
looking for and identifying trends in the data, statistical analysis of data,
identification of spurious points (i.e. data points that are outside the statistically
significant expected range of your data), the importance of significant figures,
preparation of data tables and the analysis of tabulated data, preparation of
figures, analysis of data plotted on figures, and working with numbers. While it
may look complicated at first, youll find that with practice and experience you will
become very skilled at analyzing scientific data.
Some definitions
What are data? What are results? What are conclusions? And whats the
difference?
Each number that you measure and record during the course of an
experiment is called a data point, and all of your data points taken together make
up your experimental data, also called raw data, or primary data. Data must be
compiled and analyzed to yield results, and you will analyze your results to arrive
at conclusions. For example, in the context of our refrigerator problem, the
individual temperature measurements are data points, and all of the
measurements taken together are your experimental data. Your results are the
quantitative values derived from the analysis of your data. The results of your
refrigerator experiment will include the time required for the refrigerator to reach
constant temperature at each setting and the value of the equilibrium
temperature at each setting, as well as an analysis of the influence of the position
of the thermometer on the equilibration time and equilibrium temperature.
Finally, your conclusions will consist of a statement or statements of what you
learned from your results and how you can use that information, such as whether
or not the temperature is uniform within the refrigerator (hence where is the best
place to put different foods), etc.

32

How good are your data? - precision vs. accuracy, and reproducibility
Precision and accuracy are terms that are used to describe the quality of
experimental data. If repeat measurements on the same sample using the same
instrument yield the same value, we say that both the instrument and the
measurements have high precision. If the measured value is close to the true
(or correct) value of the property being measured, we say it is accurate, or has
a high accuracy (note - we will revisit the concept of the true value below, in the
section on random error). Note then that it is possible to make measurements
that are precise but inaccurate (how?), or accurate but not precise (how?).
Reproducibility is concerned with the actual product or experimental
sample being investigated, such as the product of a synthesis reaction, a solution
or mixture, a new material, etc. A product is reproducible if it can be made up at
different times by different people and still exhibit the same properties. If a
product is not reproducible, that is, if duplicate batches do not yield reproducible
results, then it is necessary to review the methods used to produce it so as to
ensure better control over its properties. Factors that may contribute to a product
not being reproducible include human factors (such as inconsistent technique on
the part of the operator or operators), the use of equipment that is not suited to
the product (for example, a sample with a high viscosity may require specialized
mixing equipment to ensure complete mixing), or problems with the process
itself, such as incomplete mixing, poor control of the temperature during
processing, not allowing sufficient time for a system to equilibrate, etc. The
solution to the problem may range from replacing torn insulation to redesigning
the procedure.
How bad are your data? - random vs. systematic error, experimental error, and
plain old mistakes
Random error occurs due to natural small-scale fluctuations in your
sample itself or in physical factors that affect the operation of your instrument,
hence its output or readings. All systems are subject to some natural fluctuations
in their properties over short time scales. While this fact of nature cannot be
erased, there are ways to ensure that random error in your measurements is kept
to a minimum. For example, suppose certain properties of your experimental
sample are sensitive to temperature. You can minimize random fluctuations in
these properties by using a high-precision temperature controller, i.e. one that
cycles on and off within a sufficiently narrow range about the set point that your
sample will not be affected. How narrow is sufficiently narrow? Ah, theres the
question. You have to decide that based on what you know about the sample,
what temperature controllers are available (within your budget), and how much
random error you can tolerate in your results! As a second example, suppose
your instrument is sensitive to vibrations. Anything that causes such vibrations (a
truck rolling by, a loud bang, air currents, etc.) can affect the operation of your
instrument, so some of your measurements will likely be inaccurate because they

33

will include some error of indeterminate value arising from randomly-occurring

vibrations. In this case it would be wise to isolate the instrument from external
vibrations by placing it on a free-floating vibration table (look it up!) and to isolate
it from wind currents by surrounding it with PlexiglasR, for example. In general, in
order to minimize random error it is necessary first to identify the factors to which
your experimental sample and/or your instrument are sensitive, and then to try to
control them to the greatest extent possible. Note that sometimes it is not
possible to isolate all the causes of random error, and you must simply accept
that some variability in your results is natural and unavoidable. By the way, note
the use of accuracy and precision in the above discussion. Do you see the
difference?
Systematic error occurs when there is a consistent flaw in the experiment.
This can take the form of an undetected problem with the instrument, a problem
in ensuring the validity of your assumptions, or operator error where the operator
performs the experiment with the same improper technique each time. Note that
this is not the same thing as the operator making random mistakes! Examples of
problems with the instrument include improper calibration, an undetected liquid or
gas leak, imperfect insulation, a component that is out of alignment, etc. The
resulting error in your measurements may either be the same for all your
experiments, as in the case where the instrument is not properly calibrated and
the baseline is off by a constant amount, or systematically (as opposed to
randomly look it up!) varying with experimental conditions, as in the case of
imperfect insulation, where the resulting heat leak may not cause much error
when the experiment is run at or near ambient temperature but will cause an
error of increasing magnitude with increasing temperature of the system.
Examples of assumptions that may be difficult to ensure include the assumption
that a system is at equilibrium or reaches equilibrium rapidly, the assumption of
laminar flow or turbulent flow throughout the system, the assumption that a
system is everywhere well mixed, etc. Examples of consistent operator error
include inaccurate readings due to parallax error (look it up!), inconsistent human
reaction times, personal bias in observations, etc. Systematic error leads to
inaccurate results, i.e. results that differ from the true value or trend. Hence it is
important to evaluate your results in light of your expectations or hypothesis in
order to detect hidden systematic errors (or incorrect expectations!).
Because of the random error inherent in any measurement or sample, as
well as the possibility of undetected or uncorrected systematic error in every
experiment, repeat measurements on identical systems do not generally yield
identical values. Rather, it is normal to find that repeat measurements yield
different values that nevertheless stay within fixed limits. This range of values
defines the experimental error (also termed expected range) for your system,
which is a measure of the level of accuracy of your results (remember what
accuracy means?). It is often expressed in terms of a mean value +/- the
standard deviation (see section on Statistical Analysis), with the standard
deviation expressed either as an absolute number or as a percent of the mean

34

(for example, 50 +/- 10 is equivalent to 50 +/- 20%). Once you have established
the magnitude of the experimental error for your system, you can evaluate new
data points to assess whether they lie within the expected range, or within
experimental error.
So what about plain old mistakes? Plain old mistakes are isolated (as
opposed to systematic) occurrences arising due to human error that in some way
alter the experimental sample or apparatus, hence perhaps the outcome. Thus
the results you get from an experiment where a mistake was made (i.e. where
the system was in some way different from in your other experiments) should not
be lumped together with the rest of your results. However, it is always a good
idea to review those results anyway before you discard them completely, just to
see if something can be learned from them that might suggest additional
experiments or a new way of looking at your system. Note that all experiments,
including the ones whose results you decide to discard (along with a comment
about why you discarded them), should be logged in your notebook.
How to present your data significant figures and scientific notation
Significant figures are, literally, digits that represent values you actually
know based on your measurements. All measurements carry with them some
level of uncertainty. In particular, the last digit of a reading, whether from an
analog scale (such as is typically found on thermometers, rotameters, etc.) or a
digital readout, has an uncertainty of order +/- 1 or more (think about it!
Suppose, for example, the indicator on an analog meter lies somewhere between
scale divisions 7 and 8. If two different people read the meter, isnt it likely that
one person will say, maybe, 7.3 while the other says 7.4?). Thus this level of
uncertainty should be carried through any calculations you make involving your
measured values so that the final result is expressed in such a way that it
conveys the appropriate level of uncertainty as well. What does that mean,
exactly? Some numbers illustrate the principle. Suppose your measurement
gives a value of 2.3. Now suppose the result you need is obtained by taking the
reciprocal of the measured value. The reciprocal of 2.3 may be calculated to as
many decimal places as you like. For illustration purposes lets use 4 decimal
places, so your result is 2.3-1 = 0.4347. But your readout of 2.3 actually means
that the true value lies in the range 2.2 to 2.4. So your result actually lies in the
range 2.2-1 to 2.4-1, or 0.4545 to 0.4166. What is the true result, then? The true
result is the result that encompasses all of the uncertainty in your data. Thus it is
the result of your calculations expressed to the same number of significant
figures as your data. In this example, then, your result should be given as 0.43.
Given the implied level of uncertainty of +/- 1 or more in the last digit, this value
does indeed encompass the range of actual values derived from your data, as
you can see from the numbers. Indeed, reporting your calculated result to
anything more than 2 significant digits in effect misrepresents the accuracy of
your data and of your experiment.

35

So what are the rules regarding significant figures? In general your final
result should reflect the accuracy of all of your data taken together, and should
therefore be presented with the same number of significant figures as your least
accurate measurement. There is no point to adding together numbers like, say
24 and 13.2659 and claiming that you are certain of any answer other than 37.
The digits that come after the decimal point in 13.2659, while they may have
been measured accurately, are only significant in the context of the
measurement that generated them. The value of 24 from above is understood to
carry an uncertainty of +/- 1 (or more). By reporting the sum of 24 and 13.2659
as 37, the implied uncertainty in the last digit (7) allows for the extra decimal
places as well as the uncertainty in the measured value of 24.
Scientific notation provides a convenient way to represent very large and
very small numbers and to keep track of significant figures. Using scientific
notation, every number is represented as a small number (of order 1, or 100)
times a power of 10. Thus 7,543 becomes 7.543 x 103, and 0.000627 becomes
6.27 x 10-4. Note that the base should contain significant digits only.
Bases for comparison - order of magnitude analysis, sensitivity analysis, and
efficiency
An order of magnitude analysis challenges you to distinguish quantitatively
between large and small effects in your system. For example, suppose you have
observed that there is heat loss from your apparatus, and you want to analyze
the possible cause. The possible mechanisms for heat transfer include
conduction, convection, radiation, and evaporative cooling. Can you rule any of
these possibilities out based on the known magnitude of the heat loss? To do
this, you must estimate the relative magnitudes of all possible mechanisms at
work in your system and compare them. Typically you will find that some effects
far exceed other ones in terms of their importance in your system. Note that it
may be necessary to identify appropriate governing equations for each of the
effects, and to look up or estimate the relevant material or performance
properties (such as thermal conductivity, emissivity, etc.) required to use these
equations to perform actual calculations. Alternatively, it is often possible to
compare the relative importance of different factors without knowing the actual
values for these constants, if you know or can estimate their order of magnitude
(that is, the exponent when its value is expressed in scientific notation).
If you want to know how your results are influenced by small changes in
some condition of the experiment, you must perform a sensitivity analysis. For
example, suppose the ambient temperature in your laboratory varied from 20oC
to 23oC (that is, 68oF to 74oF) during the course of your experiment, and you
suspect this might have had an influence on your results. Using appropriate
equations, compare the expected outcomes (that is, the expected values of your
measurements, y) at both ends of the observed temperature range to determine

36

the percent difference in the results, i.e. their sensitivity to temperature over the
temperature range of interest. One good way to do this is to take the ratio of the
two outcomes. In this way, any system constants in the equation will cancel out,
which means that you dont need to look up their values to perform the
comparison!
Efficiency is a measure of how close a performance characteristic of a
system comes to the ideal value. It is calculated by taking the ratio of the actual
measured performance parameter to the ideal or limiting performance parameter,
expressed as a percent. For example, for the case of an electric water pump, the
input to the device is electricity and the output is hydraulic power. If the hydraulic
power output equals the electrical power input (corrected for consistent units),
then there is no energy lost by the pump and we say it is 100% efficient.
However, since friction and heat losses are inevitable, efficiencies <100% are
expected. As a second example, consider a multistage separation device such
as a distillation column. The maximum extent of separation at each stage is
limited by the equilibrium composition of the vapor and liquid at the temperature
and pressure of the stage. Thus the column is 100% efficient only if the system
reaches equilibrium at every stage. However, this is unlikely because the holdup
time in each stage is relatively short; hence the efficiency of a distillation column
is typically <100%. It is good engineering practice to monitor the efficiency of
every process and piece of equipment and attempt to maximize these
efficiencies to the greatest extent possible within the constraints of the system
(note these constraints may range from requirements placed on the rate of
production to cost constraints to availability of equipment, etc.).
Statistical Analysis
Statistics is the branch of mathematics that is concerned with the behavior
of large numbers of similar samples. Given that random fluctuations in a
systems properties are an unavoidable fact of nature, no single measurement on
a sample can be used to predict the results of future measurements on other
identical samples. However, if measurements are made on a large number of
samples, you will obtain a range of values for a given property. This range is in
fact a more correct representation of the expected value than the result of any
single measurement. Statistics provides a means to analyze sets of data in
terms of their expected value and expected range. Below are brief descriptions
of two common statistical procedures for analyzing data. The first (mean,
variance, etc.) is the correct averaging procedure for analysis of multiple repeat
measurements or measurements on multiple identical samples. The second
(linear regression analysis) is a procedure for fitting a straight line through a
series of data points whose values vary with respect to an independent variable.
For a more complete discussion and sample calculations of confidence intervals,
please refer to any basic statistics textbook, such as one of those listed in the
Bibliography.

37

Mean, variance, standard deviation, and confidence intervals (student t-test)

The mean is the average value of a set of numbers. The variance and the
standard deviation provide measures of the spread of values about the mean.
The variance is the average difference squared between the actual data points
and the mean value. The standard deviation is the square root of the variance,
hence it is the average distance between the data points and the mean value.
Results are typically reported as the mean +/- standard deviation, with the
number of data points, N, indicated in parentheses (e.g. 59 +/- 6 (N=5)). The
larger the sample (hence the larger the N), the more reliable the results.
However, given that there is almost always some spread in the values of your
measurements, even on seemingly identical samples, how can you be sure that
a single measurement lies close to the mean? If the values can be assumed to
lie within a normal (i.e. bell-shaped curve) distribution about the true mean, then
we can define a range of values within which we can be confident that most
measured values will lie. Here again, the higher the value of N, the higher the
level of confidence in this estimated range. This confidence interval may be
calculated using a method called the Student t test.
Linear regression analysis, standard error of estimate, and correlation
coefficients
Suppose you find that your output variable y varies systematically with
input x (or time). If you can identify an algebraic equation that describes this
variation, then you can use this equation to determine the expected value of the
output y for any arbitrary value of input x without having to conduct additional
experiments. The simplest algebraic functional form is a first order polynomial,
that is, the equation of a straight line (y = mx + b, where m is the slope and b is
the y-intercept). While a straight line is not always the appropriate curve to fit
through a set of data, it is often a good first approximation. The procedure for
fitting a straight line through a set of data points is called linear regression
analysis or linear least squares analysis. It consists of locating the so-called
best fit straight line, defined as a line passing through the data points and
located such that the average distance between the line and all data points lying
above it is exactly equal to the average distance from the line to all points lying
below it. In essence, then, linear regression analysis is like an averaging
procedure, but instead of the points (y) being distributed evenly about a single
value (yavg) they are distributed evenly above and below a straight line. Now lets
extend this analogy to find the equivalent of the standard deviation about the
fitted line. This equivalent is called the standard error of estimate. It is
calculated in the same units as the y variable, and is a measure of the spread of
the data points about the regression line. Thus it provides an estimate of the
expected range of y for any arbitrary value of x. The correlation coefficient, ,
tells the extent to which the data deviate from a perfect linear correlation. Its
value ranges from 0 to 1.0 (or 0 to 1.0 for a regression line with a negative

38

slope). A correlation coefficient of 0 means the y values do not vary significantly

with x (i.e. y is invariant with x, within experimental error). A perfect correlation,
that is, one in which every data point lies exactly on the regression line, has a
correlation coefficient of 1.0 (or 1.0 if the slope is negative). The higher the
degree of scatter, that is, the higher the uncertainty in the slope, the more
deviates from unity (or 1.0).
Most scientific calculators today have hard-wired linear regression
analysis features that simply require you to enter the x and y coordinates of your
data points and then give you the slope and y-intercept of the best-fit regression
line (read your calculators manual to find out how to use this feature), as well as
the correlation coefficient. Unfortunately, calculation of the standard error of
estimate is not generally included among the standard features of most
calculators. For details on the calculation procedure for the best fit straight line
and the standard error of estimate, refer to any of the statistics texts in the
Bibliography.
Non-linear regression analysis, semi-log plots, log-log plots, and the F test
For data that do not lie on a straight line, it is possible to fit higher-order
polynomials or other non-linear functions such as exponentials, trigonometric
functions, etc. This is computationally more complex than linear regression and
also requires large numbers of data points to be reliable. There are, however,
three cases where non-linear data may be fitted relatively easily using linear
regression analysis. The first is the case of a function that may be subdivided
into 2 or more approximately linear regions, each characterized by a different
slope and intercept, as shown below. In this case it is possible to analyze
appropriate subsets of the data using linear regression, and to use the resulting
regression lines to estimate the expected value of points within a well-defined
range.

The second is the case where y varies exponentially with x, that is, where y=kex
(or y=ke-x), where k is a constant. This equation may be rewritten in terms of
natural logarithms, as follows: ln y = ln k + x, or x = ln y/k. If you were to plot the
data points (x,y) on semilog coordinates (where the vertical scale is a log scale
and the horizontal scale is linear), which is equivalent to plotting ln y/k vs x, these

39

data would lie along a straight line which could be fit using linear regression
analysis. The inputs for the linear regression are x and ln y, and the output
intercept will correspond to ln k (work this out in the margin and see how it works!
What is the value of the slope?). Finally, the third case is where y=kxa, where k
and a are constants. Then log y = log k + a log x, and a plot of log y vs. log x will
produce a straight line of slope a (note the intercept has no meaning in a loglog plot (why?)).
For cases where the data are clearly non-linear and do not lie on a straight
line on either semilog or log-log coordinates, it may be desirable to find the bestfit polynomial passing through the points. This may be accomplished
numerically, but in order for the fit to be reliable it is necessary to have a large
number of data points. Then it is necessary to establish what is the highest order
of the best fit polynomial that is still statistically significant. To illustrate this, try
plotting a second order polynomial on a piece of paper. Now plot 3rd and 4th
order polynomials. You can see that at least over some portions of these curves
the shapes look more or less the same to the eye. How, then, can you decide if
your data are modeled more accurately by, say, a cubic than a quadratic
equation? This question is answered using the F test. The F test defines a
criterion for deciding what is the highest statistically significant order of a
polynomial fitted to a set of data. Details on non-linear curve fitting calculations
are beyond the scope of this manual. Briefly, a polynomial of order a is fit to a
set of data and the criterion given by the F test (note F test criteria are
tabulated in most statistics books) is applied to see whether the order of this
polynomial is too high to be considered significant based on the goodness of the
fit and the number of data points used. Note that the F test can even be used to
evaluate whether the results of a linear regression analysis are significant, that is,
whether the variation in y(x) is due to a real dependence of y on x (i.e. y(x) is a
polynomial of order 1) or simply due to random error (i.e. y is invariant with x, that
is, y(x) is a polynomial of order 0).
Analysis of Data: Tables and Figures
Your primary data will be recorded in your laboratory notebook, as
described in the section on The Laboratory Notebook, above. Now its time to
inspect, compile, plot, and analyze those data to see what can be learned from
your experiments. Below are some rules and strategies for extracting the most
information from your data.
1 - Set up your data tables in a way that makes sense.
For your primary data, set up your table as discussed in the section on
The Laboratory Notebook. You may also wish to tabulate subsets of your data or
results. Here again, set up your tables in a way that makes sense. Make sure
the runs are cross-referenced from table to table, so it is clear which entries refer
to the same experiment. In general, shorter tables are easier to read and

40

understand than longer ones, so its often better to work off a few short tables
than one very extensive one. Write a clear title for each table. The title should
describe what is to be learned from reading the table, and should not simply
restate the column headings.
For example, in the case of the refrigerator problem, your raw
data in your notebook will include temperature readings from each
position at each setting at several different times, finishing when repeat
readings hour apart give the same temperature. You may wish to
prepare a smaller summary table with a row for each temperature setting
and columns for the endpoints, that is, time to equilibrium and the
equilibrium temperature, at each of the positions, as follows:
Table Refrige1. Effect of thermometer position on time to equilibrium and
equilibrium temperature inside the refrigerator.
INDICATOR
SETTING

POSITION 1
time to eq.
(min)

temp. at
o
eq. ( C)

POSITION 2
time to eq.
(min)

temp. at
o
eq. ( C)

POSITION 3
time to eq.
(min)

temp. at
o
eq. ( C)

1
2
3
4
5
6

Ultimately each of these smaller summary tables may provide the basis for a plot
as well.
2 - Make sure you examine all raw numbers first.
Although you will be using some of your numbers to calculate other
numbers, it is important to analyze your raw numbers first. This will help you to
detect invariants (i.e. constant values), trends, and systematic errors, if any, and
may also provide physical insight into the system.
Start by looking critically over all of your raw data. Do the numbers look
right? Do they appear to follow a trend? Do any points stand out as seeming too
high or too low? If so, are there any comments in your comments column that
could explain these anomalies? Now make sure that the same comments did not
also apply to points you do not suspect! Below are some examples of things you

41

should be on the lookout for in inspecting your raw data, and how to handle
them:
Invariants, or systematic constants - Some values are expected to be invariant
throughout the experiment, such as ambient conditions, conditions (such as flow
rate, temperature, pressure) of an input stream, pressure upstream from a pump,
electrical power draw of a piece of equipment, etc. Always inspect your data to
see if these values actually did remain constant. If they didnt, look in your
comments column for possible explanations and determine what your course of
action should be. You may decide that its a good idea to repeat some or all of
the runs, to perform additional experiments under different conditions, to
redesign the experiment, to fix the equipment, or you may find that some
interesting phenomenon you hadnt previously thought of can account for the
unexpected variability in your measurements.
Scatter As noted above, some random scatter is expected in all experimental
measurements. However, look over your data to assess whether the variations
you observe are within expected or acceptable limits. You may decide that the
level of scatter in your data is so high that its not possible to draw reliable
conclusions. In that case its probably a good idea to repeat or redesign some or
all of your experiments in order to reduce scatter.
Trends - Inspect sets of data that you expect should follow a trend, such as a
change (up or down) in output y as you increase input x. Do you see the trend
you expected? How do you know if the trend is real? Suppose, for example, that
you found that y went up steadily from 23 to 27 as you increased x. Now
suppose you also found that the standard deviation of repeat measurements of y
is +/- 10%. This means that the difference between 23 and 27 is not statistically
significant (how do I know? Calculate the range of values that is implicit in each
of these two numbers and youll see23 +/- 10% means anything from 20.7 to
25.3. Furthermore, 27 +/- 10% means anything from 24.3 to 29.7. Since these
ranges overlap each other, we say that the difference between 23 and 27 is not
statistically significant, or that these two values are the same within
experimental error.). Thus it is not correct to conclude that you observed a trend
in y vs x, though you may be right in expecting that there is such a trend. What
to do? Try performing additional experiments over a wider range of x. If the
trend is real, y values measured at higher x will be significantly different from
those measured at lower x, within experimental error. For example, given the +/10% standard deviation of measurements in this example, a y value of 29 is
significantly different from a y value of 23, and may indeed be observed at a
higher value of x than you studied in your first set of experiments.
Peaks or minima Do your raw data reveal any peaks (or minima) in y? If so, it
may be a good idea to conduct additional experiments at x values in the vicinity
of the peak (or minimum) in order to establish that the peak (or minimum) is real
and to determine its position and magnitude more accurately. For example,
suppose your measurements gave y values of 2, 3, 12, 6, and 3 at x values of

42

10, 20, 30, 40, and 50, respectively (see below). Is there a peak in this function?
Probably. Where does the peak lie? Actually, you cant tell! The highest y value
you measured lies at x=30, y=12. But does that point represent the peak itself,
or does it lie to the left or the right of the peak? On the curve below, connect all
the diamonds (series 1 points) in pencil using a solid line. Now pencil in the
following points as an illustration. Using circles as the symbol, plot points at
(35,14) and (45,9) and connect all the points (series 1 diamonds plus circles)
using a dashed line. Now using squares, pencil in points at (35,9) and (45,13)
and connect the diamonds plus squares (omitting the circles) using a dotted line.
Now answer this where does the peak in y vs x lie? The answer is, your
original data do not tell the whole story. Only by measuring y at intermediate
values of x (x=35 and x=45) can you really determine the shape of the curve and
accurately locate the peak!
Sample curve of y vs. x data showing evidence of a peak

Transition points In addition to peaks or minima, look for points where there is
an abrupt change in the slope of y vs. x. Other landmarks to look out four include
any step change in y(x), any point where y levels off with respect to x, or in
general any point where the trend in y vs. x is not smooth. Note the coordinates
of these transition points, and then look to see if there are transitions in other
properties you may have measured occurring at or near the same conditions.
Spurious points Spurious points are data points that do not lie on a trend line or
within the expected value of a measurement. If you detect such a point, you
should question whether it is real or in some way invalid. Ask yourself the
following questions. Where was the measurement taken? Whats happening at
that point in the system? Whats upstream of that point? Whats downstream?
What was happening at the time you took the measurement? Transient factors
such as power surges, temperature fluctuations, or even someone bumping into

43

the instrument may contribute to erroneous measurements, and should be

considered as possible explanations. If you suspect that a point is invalid, you
should apply the student t-test (see above) and/or repeat the experiment, and
then decide whether or not to disregard the point in your data analysis. You
should also try to identify the cause for the discrepancy so as to learn more about
your system.
As an example, in the case of our refrigerator problem, suppose the time to
equilibrium was the same for every setting except setting number 3, where the
equilibration time you measured was significantly longer. Look in your Comments
column for possible explanations. Consider possibilities such as human error
(maybe the refrigerator door wasnt closed properly), environmental variations
(maybe someone put a container of hot soup into the fridge when you werent
looking!), electrical or mechanical malfunctions (was there a transient power
outage (sometimes called a brownout) during your experiment?), etc. If you
cant rule out these possibilities, then you should repeat the experiments on
setting number 3 before making any determination as to the validity of your result.

3 Look at derived values (e.g. deltaP = P2-P1, etc.) next

As noted above in the section on The Laboratory Notebook, your data
tables should include columns for calculated results, such as deltaP, deltaT, heat
transfer coefficient, etc., that are derived from your measured values. For each
experiment, think carefully about what calculations to perform. Make sure to use
the correct units. Note all relevant equations and unit conversions in footnotes
below your table. Inspect the calculated results for the same factors you looked
for in your raw data, as itemized above. Here again, be on the lookout for
unexpected points or trends, interesting behavior, etc.
4 - Decide how to plot your data.
There is a wealth of information contained in your data, much of which can
only be detected from appropriate plots of your data points and results. For
example, suppose you measured some quantity y at different times. A plot of
your data points as a function of time will reveal information such as the
functional form of y vs. time (such as linear, approaching an asymptote, etc.),
what the time rate of change of y is, and what the extrapolated value is for y at
time t=0. It is important to become skilled at preparing plots in such a way that all
of the information contained in the data is revealed.
In plotting your data, some common sense rules apply, as follows:
Set reasonable scales for your x and y axes. Your x and y ranges should reflect
your actual numbers. Theres no sense setting the y scale to go all the way to
100 when your highest y value is 50. Note that its also not necessary to start at
0 if your values start much higher. In general you should plan your plots such
that your data fill the entire field.

44

Use reasonable scale increments and grid lines. It makes no sense to use grid
lines spaced 3 increments apart on your x axis when your measurements were
taken at x values of 5, 10, 15, etc! Be particularly careful when preparing plots
using the computer the computers default settings are not necessarily the ones
that make the most sense based on the numbers.
Label all axes, with units included. Also, your scale markings should generally
be written in scientific notation with no more than 3 significant figures, for clarity
purposes. How can you do this if your x or y values are large numbers? Heres
an example. Suppose you want to plot viscosity (in centipoises, or cP) on your y
axis, and your measured values for viscosity lie in the range 0-40,000 cP.
Suppose also that you wish to tick off scale divisions every 500 cP. Mark off the
scale divisions on your y axis as 5.00, 10.0, 15.0, 20.0, etc. Now in the axis label
write one of the following: viscosity [cP] (hundreds) this indicates that the
numbers you read on the scale represent the viscosity in hundreds of cP;
viscosity x 10-2 [cP] if viscosity x 10-2 is equal to 5.00 cP, say, then viscosity is
equal to 500 cP; or viscosity [cP] (x102) putting the x102 in parentheses says
that the scale readings are meant to be multiplied by 102 (that is, viscosity is
actually equal to the scale reading times 102).
Show error bars and define what they represent. When plotting points
representing average values of multiple repeat measurements or of
measurements taken on multiple identical samples, always show the mean value
(plotted as the data point) as well as error bars to show the standard deviation
about the mean. The size of the error bars provides a clear picture of the
reproducibility of your samples or your experiments. In addition, since error bars
span the range of possible values for each data point, they also show the
magnitude of your experimental error and may be used to assess whether the
difference between two values is statistically significant (data points with
overlapping error bars are said to be not significantly different within
experimental error). Note that you should always indicate in the caption or
somewhere else on your figures what the error bars represent (i.e. repeat runs
on the same system, different samples, different operators, etc.) as well as the
sample size (N).
Use symbols in a way that makes your plots easy to read and cross-reference.
The idea here is to let the symbols provide a visual cue that directs you to
compare sets of data that have something in common. Here are some tips. 1 Always use the same symbol for the same parameter on different plots. For
example, say you measured 2 different properties of a system over time at 3
different temperatures, T1, T2, and T3. You will need to prepare two plots one
of property 1 vs. time with temperature as the parameter (i.e., with a different
symbol for each of the 3 temperatures), and a similar one for property 2. It
makes sense to use the same symbol (e.g., circles) for the T1 data on both plots,
and, say, squares for all T2 data and triangles for T3. This makes it easy to
compare corresponding sets of data on multiple plots (sketch these 2 plots in the

45

margin to see what I mean). The same rule applies to control data use the
same symbol to represent data from control experiments on every plot. 2 Use
open and closed symbols with the same shape for corresponding sets of data, as
warranted. For example, suppose you measured some property of two different
systems, and for each system you made measurements both with and without
salt (for a total of 4 sets of data). You could choose one shape (say circles) for
the symbols representing system 1 and a different shape (say squares) for
system 2, but use closed symbols for with salt and open symbols for without
salt. If you plot all 4 sets of data on the same set of axes, your eye will
immediately be drawn to compare the circles (system 1 with and without salt), the
squares (system 2 with and without salt), the closed symbols (both systems with
salt) and the open symbols (both systems without salt). Thus the symbols direct
you to compare related sets of data. (Try sketching a plot with 4 sets of data as
described to see what I mean.). 3 Make sure that the symbols are large
enough to be visible, and will appear in a black-and-white photocopy of your
plots. Your experimental data points are really the only factual information
presented on your plots; thus it is imperative that they are visible and big enough
that the different symbols are easily distinguished from one another. You may
elect to draw curves through these points, but the data points are the only points
you actually know to be true. Printouts of figures are often reduced for
publication purposes, so make sure that the data points are big enough that the
different shapes are distinguishable even if the figure size is reduced. If you
intend to project these figures on a screen during a presentation, the points must
be large enough that the different shapes can be distinguished from one another
even from the back of the room. Remember that if you use a color printer to
produce a plot from the computer, yellows, light blues, and some other colors
dont appear on photocopies of those plots.
The legend should list symbols from top to bottom in the order in which they
appear on the graph. The legend is the list defining the meaning of the symbols
used to represent the data points. In the legend the symbols should be listed in a
logical order, preferably one that aids the eye in reading the plot. Thus if the data
represented by circles consistently lie above the data represented by triangles,
then when the symbols are defined in the legend list the circles above the
triangles, not the other way around! This makes the figure much easier to read
and interpret, as you can see from the example below.

46

Note that if you have 3 or more symbols corresponding to parameters that have
numerical values (such as 3 different temperatures), it generally makes sense to
list them in the legend in either ascending or descending order, even if their
corresponding data sets do not lie in order on the plot. Either way, give the
matter some thought and try to make the plot as easy to follow as possible. Here
again, if you use a computer to generate your plots, dont let the computer make
the decision for you as to how the figure should look!
If you make multiple plots over the same range of x, use the same horizontal
scale for all the plots. That is, use the same actual distance (in inches or cm)
between equivalent scale divisions on your x scale. This makes it easy to
compare the behavior shown on different plots. Similarly, use the same vertical
scale for all graphs with the same y, if appropriate
In plotting your data, try to group together experiments that as a group can tell a
story. For example, suppose you measured the effect of concentration of solute
on the viscosity of a solution at 3 different solution temperatures. In this system
there are 2 operator-adjusted variables, namely, concentration and temperature,
and 1 output, namely, viscosity. You could prepare a separate plot of viscosity
vs. concentration for each temperature, but these plots would tell you nothing
about the effects of temperature on the system. It is better to plot as much data
as possible on the same set of axes in such a way as to reveal as much
information as possible about the system on one plot. For this example, there
are 2 good ways to represent all of the data on a single plot, and each way tells a
different part of the story. First, you could show the effect of concentration on
solution viscosity at each of the 3 temperatures by plotting viscosity (y) vs.
concentration (x), with temperature as the parameter (z). Alternatively, you could
prepare a cross-plot (that is, reversing input parameters x and z) showing the
effect of temperature (x) on viscosity (y) for each of the concentrations (z) you
studied. In the space provided below, sketch each of these two alternatives side
by side, labeling the axes and showing the expected behavior of the curves.

47

Just as above in the discussion of data tables, it is often necessary to

prepare a series of plots in order to represent your data completely. Generally it
is a good idea to draw plots by hand directly in your lab notebooks initially, to
help identify trends and/or spurious points or unexpected behavior (as discussed
above) in the data, or suggest additional experiments that may be needed, e.g.,
to help locate a peak or minimum in the data. The plots are usually based on the
data tables and summary tables described above, and should be designed to
answer the questions you sought to address when you planned your
experiments. Below is a detailed discussion of how to plot and analyze the data
from the refrigerator problem. As you read through this discussion, sketch the
plots that are described in the margin or on a separate piece of paper, and also
look in the section on The Laboratory Report for sample plots:
In the case of the refrigerator problem, you set out to learn 1
whether the temperature within the refrigerator depends on position, 2
what is the equilibrium temperature at each of the 6 settings and 3
positions, and 3 - how long the temperature takes to stabilize at each
position after the setting is increased by 1 unit. Your first task is to prepare
plots of all your temperature data (y) vs. time (x) with position as the
parameter (z) (that is, using a different symbol for each position and
connecting only the points with a common symbol) for each of the 6
temperature settings (6 plots total). From these plots you will obtain
quantitative results for the time rate of change of temperature, the
equilibrium temperature, and the equilibration time for each setting and
position within the refrigerator, and you will also be able to tell immediately
whether or not these quantities vary with position. You may also choose to
prepare cross-plots, 3 in all (1 for each position), of temperature (y) vs. time
(x) with setting as the parameter z. These cross-plots will allow you to
compare the rate of change of temperature and the equilibration times for
each setting at each position.
Now answer question 1 does position play a role? Look at your
first set of 6 data plots showing position as a parameter for each of the 6
settings and decide whether or not any differences you observe are
significant within experimental error (note you may need to perform some
repeat experiments to determine the range of experimental error in your
system). If you determine that the temperature does not vary with position
in the refrigerator, then you should state this conclusion in any report on
your experiments but you do not have to consider position as a factor in
the rest of your analysis. However, you should now average together the
temperature readings from all 3 positions at each setting and time to give a
complete picture of the system. In addition you should prepare a new plot
of temperature (x) vs. time (y) with setting as the parameter (z) and showing
error bars for each data point reflecting the average reading +/- standard
deviation over all 3 positions at each setting and time. Note that this
process has enabled you to represent all your data on a single plot, rather
than the 9 plots described in the preceding paragraph! On the other hand,
if your data show that position is a factor, then you should include a
discussion of the differences between the positions (Hot spot near the
lightbulb? Slow equilibration time near the door? Cold spot in the back
corner?) in your analysis of your results, and make sure in the rest of your
data analysis that you consider each position as a separate case. Note that
you may find, for example, that the final equilibrium temperature is the

48

same for all positions but that at every setting it takes longer to reach
equilibrium at position 3 (near the light bulb) than at position 2 (in the
center of the fridge). This is an important result and directly addresses
part of your initial motivation in performing these experiments!
On to question 2 what is the equilibrium temperature as a function
of setting and position? Note that we have now moved from analyzing data
to analyzing results (remember the difference?)! The equilibrium
temperature at each setting can be read off your temperature vs. time plots
(in the context of these experiments, what is meant by equilibrium
temperature? What is meant by equilibration time? More on that
later). Review all of these plots and then prepare a table, such as Table
xxx (shown above) listing setting number in the leftmost column and
equilibrium temperature and equilibration time for each of the 3
positions. Then prepare a plot of equilibrium temperature (y) as a function
of setting (x) with position as the parameter (z). Note that if you concluded
that position is not a factor, this plot does not require the use of setting as
a parameter at all, but should instead depict the mean equilibrium
temperature (with error bars spanning +/- standard deviation) at each
setting obtained by averaging together the values measured at all 3
positions. Here again weve extracted information from 9 data plots and
presented it on a single results plot! Pretty good!
Now for question 3 what is the equilibration time as a function of
temperature and position? Note that this question also refers to your
results rather than your primary data. The equilibration time can be
determined from the data plots, above. However, you must define what you
mean by equilibration time in terms of those plots. For example, you may
choose to draw smooth curves through the temperature vs. time data for
each setting and interpolate between actual data points to estimate the
time at which the temperature reached its equilibrium value. Alternatively,
if the intervals between your temperature measurements were sufficiently
short, you may simply choose to use as the equilibration time the time
corresponding to the first time you measured the equilibrium temperature
at each setting and position. Now prepare a plot of equilibration time (y) as
a function of setting (x), either with position as a parameter (z) or with error
bars covering the range of measurements at all 3 positions. Remember
when you ultimately write up your description and analysis of this plot you
must explain exactly what you mean by equilibration time in terms of your
original data plots.

Below are some questions you should address in preparing plots of experimental
data and results:
Which variable should I plot on the x axis? Which variable should I plot on the y
axis? In general (though there are exceptions to this rule) you should plot your
data with the input value as the independent (i.e. x) variable and the output
values (data) as the dependent (i.e. y) variable. Note that in this context the
input variable refers to the value that you set before the experiment, and the
output variable refers to the value that you measured in the course of the
experiment. Note that time, or more accurately, elapsed time, is also
considered an independent variable (why?).

49

What data should I plot together on a single set of axes? If you conducted repeat
experiments under conditions that were identical except for the value of one input
parameter (such as temperature, flow rate, etc.), then its a good idea to plot the
data from all these experiments on the same set of axes, using a different symbol
for each of the different values of the input parameter z.
When is it appropriate to use two y axes? If you measured 2 different outputs as
a function of the same input value (or as a function of time), its often instructive
to plot both sets of output data together on the same figure. This really amounts
to superimposing two plots (of two different outputs, y1 and y2), on the same x
axis. It is accomplished by using two y axes, one on the left for property y1 and
another on the right for property y2. The scales on the two y axes in general are
unrelated to one another and have different units, but the x axis is the same for
both properties. Different symbols should be used for y1 and y2, and there
should be an arrow from the symbols (or curves drawn through the symbols)
pointing to the appropriate y axis. Plots with 2 y axes are useful for determining,
e.g., whether a transition occurs at the same x value for 2 different properties, y1
and y2.
What results can I obtain from the plots of my data? How should I compile and
make sense of those results? Plots of data can reveal results such as endpoints
(x or y at equilibrium), slopes of y vs. x, positions of landmarks such as peaks,
minima, or transition points, etc. Inspect each individual set of data for such
results, then tabulate the results, and then prepare additional plots showing the
influence of your input parameters on these results. Now inspect these new
tables and plots to see what you can learn! Does the magnitude of the endpoint
depend on the conditions of the system? Do transitions occur in any of the
properties you measured? If there are transitions in more than one of the
properties you measured, do they always occur under the same input conditions
(x)?
Making sense of your results
Once you have completed all your calculations and tabulated and plotted
all your data and results, you are ready to think about possible explanations for
the behavior you observed in your system. First think through what you expected
to observe based on what you already know about the chemistry or physics of
the system. Then decide whether your results are consistent with these
expectations. If so, then you can conclude with reasonable certainty that the
assumptions you made in setting your hypothesis are correct. If not, you should
review all your assumptions and conduct a sensitivity analysis on your results, if
appropriate. For example, suppose you based your hypothesis on the
assumption that the temperature is constant throughout the experiment. Now
suppose you are aware that in fact your temperature controller maintains the
system at a temperature that is +/- 2oC from the set point. What would be the
effect of a difference of 2oC on the properties of your system? Try to estimate

50

the range of expected values for your results over that temperature range. If this
uncertainty in the temperature can lead to the observed deviation from your
expected result, then you can conclude that temperature is at least one factor
that might have contributed to your unexpected results. If, on the other hand,
you find that your results are not sensitive to temperature over that range, that is,
that the change in your measured values resulting from a 2oC change in
temperature is not statistically significant (i.e. within experimental error), then you
can conclude that your unexpected result cannot be attributed to the uncertainty
in the set temperature, and you must evaluate other factors to try to explain your
results.
More on the interpretation of results can be found in the chapter on The
Laboratory Report.

51

Chapter 6: THE LABORATORY REPORT

The laboratory report is a finished document that describes your
objectives, experimental procedure, results, interpretation of those results, and
conclusions and recommendations. Depending on who the intended audience is,
this document may be very extensive, covering most or all of the details of your
work, or rather abbreviated, just discussing highlights. Either way, the report
must present a faithful and accurate description of what you did, what you found,
and what the significance of this work is. In this chapter we will cover the proper
way to write lab reports, paying particular attention to writing up the discussion of
your results, as this discussion is what leads to your conclusions about the
significance of your work and is therefore the culmination of all your efforts. Note
that the study questions related to each experiment in this manual are intended
to serve as prompts for information that should go into the final lab report. This
chapter also includes a section on good writing style, with exercises to guide you
on how to write a paragraph that is clear and grammatically correct. And we will
present several sample laboratory reports, one long version and several short
versions, for the refrigerator problem.

I. What are the different sections of a laboratory report, and what goes in
each section?
Below is an outline of a complete (long version) laboratory report. A short
version will contain some of these elements, selected according to the points you
need to make and the specific interests of your audience.
Title a descriptive title that indicates the purpose and/or findings of
the experiment (Calibration of a.instrument for measuring..,
Effect of x on y, Relationship between x and y in a .unit, etc. ).
Abstract to be written last. The abstract is a very short summary of
what you did, why you did it, how you did it, what you found, and what
you think it means. This can only be written once you have completed all
of your data analysis, so that you can address the last 2 questions.
Numerical results should be included in the abstract. Remember that
often the abstract is the only part of the report that people read (indeed,
the abstract often serves as the complete short report), so it should touch
on all the important points in the report.
What was measured?
How?
Numerical results
What is the significance of the results?
Introduction The introduction provides background and context
about your experimental system, that is, the equipment or sample you
are studying, stating what it is, why it is important, how it compares with
other related systems, etc. This section also includes a statement of
your objectives, including what you measured and why.

52

What is the system being studied? Why was it chosen? How is

it different from/similar to other related systems? What is it used
for?
Some literature review, with complete citations (see below in the
section on References, for format).
What are your objectives in conducting this experiment? Why is
this objective of interest? What are the important variables and
parameters for this system? What values were investigated?
Why?
Experimental Section: Materials, Apparatus, and Procedure This
section should provide a complete description of the experimental system
and a summary of what you did. The materials section must include the
specific details (manufacturer, manufacturers address (city, state and/or
country), lot number, purity) of all chemicals. The apparatus section
must include specific details of all pieces of equipment, as described
below. The procedure is a description of what was measured and how it
was measured, but it is in narrative format, not step-by step same format
like the procedure section in your laboratory notebook. It should give the
range of values you used as inputs, state what was measured (i.e. what
are your outputs), state what was calculated (what are your results?),
and possibly describe the calculation procedure. This section should
provide enough information that someone in another laboratory could
replicate your experiments exactly, using the same chemicals and
instruments.
Materials State the name of each chemical you used, along
with the manufacturer, manufacturers address (city, state and/or
country), purity, and any additional processing or purification
steps you used (example: Chemical xyz, 99% pure (Steiner
Chemical Co., Hollywood, CA) was used as received).
Apparatus
Include a schematic diagram if needed
What instruments did you use? Somebody reading this
report should be able to reconstruct the exact same
experimental system you used. Therefore, this section
should include a description of all instruments, with model
numbers, manufacturers name, and manufacturers
address (example: The viscosity was measured using a
Ford Model 3050 Rotating Shear Viscometer, with lowshear accessory and accompanying software (Ford
Instruments, Los Angeles, CA)). The manufacturers
address is required so that someone unfamiliar with this
equipment can call the manufacturer and order the same
instrument for his or her own laboratory.
Procedure include a summary of what you did, and why. This
doesnt have to be a step-by-step description, unless its for
something like a chemical reaction or a non-standard purification

53

process. Rather, this is a narrative description of the experiment

(example: We conducted a total of 9 runs. Three different inlet
steam pressures, a, b, and c (give numbers here), were used. At
each steam pressure, 3 different inlet water flow rates 1, 2, and 3
(give numbers) were studied. The temperature at each inlet and
outlet was measured at 5 minute intervals using thermocouples
inserted into the fluid, and the equilibrium temperature and time
to equilibrium were recorded for each set of operating conditions.
Then the value of [some characteristic of the system] was
calculated using the following equation)
Results
This is probably the most important section of the lab report. In this
section, you present tables and figures of data and results, along with your
interpretation of them. The best way to start is to set up all your tables
and figures as detailed in Chapter 8 and then put them in a logical order
starting from your raw data and leading to your results. Then you will
present, on average, 1 paragraph for each table or figure, discussing what
is presented and what conclusions you can draw. Note that in the Results
section you generally do not present an explanation for your results (the
explanation belongs in the Discussion section). Rather, in this section you
must convince the reader that your results follow from your data. Note
tables and figures should be inserted into the text either on or immediately
after the first time they are cited.
In general it is preferable to present results in the form of figures
rather than tables. Tables should be reserved for describing what was
done, qualitative observations, and summaries of results, while
quantitative results are usually (though not always) best represented by
figures. Every table should have a descriptive title and every figure should
have a descriptive caption. In selecting the title or caption, remember that
these titles and captions serve to guide a reader as to where to find the
answer to a specific question. Thus Pressure drop across the pump for
different flow rates is not as informative as Effect of flow rate on the
pressure drop across the pump for flow rates in the range 15-100 gpm.
The format for a paragraph describing a table depends on what
information is contained in the table. In general, it should state what
information is given in the table, how it can be used, and how it relates to
other information presented in the section (see the sample lab report on
the refrigerator problem for examples).
In general, a paragraph describing a figure should follow the format
below. If you use this format, writing your Results section becomes very
straightforward! Note that figures may include printouts such as spectra
from your instruments, plots of data, or plots of results, inter alia.
Opening (or topic) sentence, stating what is shown in the figure
(Figure 1 shows the relationship between outputs y and input x
for all runs in the series. Figure 3 shows the effect of x on y for
3 values of z, viz., z1, z2, and z3. The effect of x on y is shown

54

in Figure 7.) If there are error bars shown on the figure, explain
their significance (The error bars show the standard deviation
about the mean for 3 identical samples, or The error bars reflect
the range of results obtained on repeat (N=3) measurements of
the same sample.).
Describe the horizontal trends, y vs. x, for constant z. What is
the shape of the curve(s)? If there are multiple curves, do they all
exhibit the same horizontal trend? Are there any transition points
(see chapter 8)? What type, and where do they occur? Do they
occur at the same or different values of x for each z? Examples:
The viscosity of solutions of solute A in water increases linearly
with solute concentration over the solute range 0.1% (w/w)
1.5% (w/w) for all 3 temperatures studied. Or, At low solute
concentrations, the solution viscosity is equal to that of the
solvent (control value) for all 3 temperatures studied. Beginning
at higher concentrations, however, the viscosity goes up with
solute concentration. The concentration at which this transition
occurs goes up with increasing temperature.). The description
should have enough information that you should be able to
sketch the curve from the description (try sketching curves from
each of the two descriptions you just read. Begin by drawing and
labeling the axes. Mark off the baseline (control) value on the y
axis. How many curves are plotted? Do they all have the same
shape? What is/are the shape(s) of the curves? What is the
parameter z?). The discussion of horizontal trends should
include some or all of the following elements, as warranted:
State the range of x
State general observations of y vs x for all z
Constant, increases, decreases, exhibits a peak,
exhibits a minimum, much scatter, no apparent
trends, etc.
Identify anomalous points
State range of y for each z, plus global range of y
Describe the vertical trends, y vs. z, for constant x. For
example, for a plot of viscosity (y) vs. concentration (x) for 3
different temperatures (z), you may write, For a given solute
concentration, the viscosity decreases with increasing
temperature. Or As z increases, y at constant x goes down.
General observations of y vs. z for all x
Always increasing/decreasing?
Do the curves cross (or converge or diverge?) If
so, where?
Are any curves superimposed on one another?
Which ones? Over what range?

55

Describe any other observations about the figure, such as

anomalous points, transition points, etc. (as discussed in Chapter
8).
Compare the results from each figure with those on other
figures. For example, if there is a peak in the property y1 plotted
on Figure 3 occurring at a particular value of x, is there also a
transition point in y2 (plotted on a different figure) at the same
value of x? Present these observations as you go through all the
figures. You may also want to prepare a table that compares all
these results, and write a paragraph explaining the table.
Discussion of Results - (Note the discussion may either be a
separate section or be incorporated into the Results section.)
Taken together, what do all these trends mean? The discussion
gives your explanation and interpretation of your measurements and
observations, stating what you think is the cause of each of the effects
you saw, what physical or chemical factors govern the response, etc.
You should attempt to explain each trend (y(x), y(z)) as well as each
transition, etc., and any other questions that were raised by your data.
Dont leave anything out. Note that it is perfectly reasonable to state
that you cannot explain something you saw. This shows that you
thought about it and ruled out all possible explanations that occurred to
you (you should also discuss why you ruled those explanations out.
This opens the way to a scholarly discussion of other possibilities, hence
discovery!).
It is often appropriate to include an error analysis in the
Discussion section, covering the magnitude and influence of random
error and the sensitivity of your results to random fluctuations in
experimental conditions such as ambient temperature. This should be a
quantitative analysis, if possible, and should address the question of
whether or not the error caused by this effect is statistically significant in
the context of your experiment (i.e. whether this error is greater or less
than the random error inherent in your experiment). Also, discuss the
goodness of the fit of your data to regression lines or theoretical
curves, if available.
Finally, how can these results be applied? What systems, new
technology, or problems can they address? This is where you get to
spread your creative wings. But dont just muse try to be specific as to
the value of your results in these contexts.
Conclusions this is a short section whose main focus should be on
how the results can be used, and what future work is suggested by
these results.
Recap very briefly what you found (remember that a summary
of what you did goes in the abstract).
How can you use these results? Use your imagination can
the results be applied to other systems? Explain.

56

Propose future experimental work suggested by these results,

and what questions this new work should be designed to
address.
References
List either in the order in which they appear in the text or in
alphabetical order by first author. Note reference numbers
should be inserted in the text in the context of the citation. For
example It has been found [ref. number], or, The xxx property
is defined as.[ref. number].
Format
Journal articles: Author 1 (last, first, mi., title), Author 2,
etc. year Title of the article, Journal name using official
abbreviations, volume number page numbers
Books: Author 1 (last, first, middle initial, title (Jr., III,
etc.)), Author 2, etc., Title of Book, edition number,
Publisher (city) copyright year page numbers
Article in a book: Authors of the article (last, first, middle
initial, title) in Title of book, names of editors, eds.,
Publisher (city) copyright year page numbers
Appendix
Raw data include data tables from your lab notebook
Sample calculations present all your equations in detail,
including an explanation of any assumptions and why these
equations apply to your system
Error analysis and sensitivity analysis, as discussed in Chapter
5 - Analysis of Data

II. Style, or how to write a good report

This section covers some basic grammar rules for writing clear sentences
and paragraphs in English. It also presents some exercises taking you through
the process of writing. When you start writing a sentence or paragraph, you may
find that you dont like the way it comes out at first. It is not uncommon to have
to rework and rewrite it several times before youre satisfied. Just to prove it,
reread the two sentences you just read, and now look at the attempts that I
rejected before deciding on those sentences:
There are
However, the process of writing a clear sentence or paragraph is not
always so simple. Typically you may find that
When you start writing a sentence, for example, you may find that it
doesnt come out the way you want
Good writing comes with practice. Easy editing comes with word processors!

57

Dont be surprised if it takes a few tries to get each sentence or paragraph the
way you want it!
Sentence structure
Every sentence in the English language has a subject (noun) and a
predicate (verb). The main point or idea of the sentence should be reflected in
the subject, with secondary information coming later in the sentence or in a
subsequent sentence. Thus you must decide what the main point is, which will
often depend on what question you are trying to answer. Consider the following
two sentences related to the refrigerator problem:
(i) Identical thermometers were used to measure the temperature at each of the
three positions.
(ii) The temperature was measured at each of the three positions using identical
thermometers.
Now compare. In example (i), the subject is thermometers and the verb is
were used. Thus sentence (i) tells how you did something, followed by what
you did. In example (ii), the subject is temperature and the verb is was
measured. Thus this sentence tells what you did, followed by how you did it.
Which one follows a more logical sequence? In my opinion the winner is
example (ii).
Writing Exercise 1. In the space below, write 3 grammatically correct
sentences, each one using all 3 of the following phrases: at night, the sun goes
down, the sky is dark. You may also include whatever other words you need to
make the sentences make sense, such as because, when, etc. Now compare.
You will see that each sentence focuses on a different aspect of a familiar
phenomenon. The best sentence to use depends on the context or the question
you are trying to answer.
a.______________________________________________________
________________________________________________________.
b.______________________________________________________
________________________________________________________.
c._______________________________________________________

58

Paragraph structure
A well-written paragraph consists of a topic sentence, body, and
(sometimes, but not always) a closing sentence. The topic sentence, which
opens the paragraph, tells what the paragraph is about. And the paragraph
should stick to that topic! Imagine you are a reader who is skimming a report
looking for the answer to a specific question. You should be able to identify
which paragraph contains your answer just by reading the first sentence of each
paragraph. The body of the paragraph is a detailed discussion of the topic.
Then the closing tells the conclusion or main message of the paragraph; it also
sometimes leads into the topic of the next paragraph. Consider these examples:
For a paragraph in your Introduction section:
Paragraph 1. topic sentence A pump is a piece of equipment used to move a
fluid against gravity. This topic sentence is general to all pumps. In a paragraph
with this topic sentence a reader would expect to find a discussion of different
types of pumps, including a comparison of how each type works, what their
different uses are, and the relative advantages or disadvantages of each, plus
other information about pumps in general. This paragraph might conclude by
saying In this study we investigated the performance of a XXX Model centrifugal
pump, or In this study we compared the hydraulic power output of three different
types of pumps, namely, A, B, and C.
This leads into:
Paragraph 2. The topic sentence should take its message from the closing of
paragraph 1. For example, The XXX Model centrifugal pump is a device
that Go on to describe in more detail information such as what the
specifications of this particular device are, what the inputs and outputs are (giving
numerical values, if available and appropriate), what it is used for, and what
nonidealities may influence its efficiency. Conclude with In this investigation we
measured y1 and y2 in order to determine.about the XXX Model centrifugal
pump. This leads logically into the Experimental Section.
For a paragraph in the Results section:
topic sentence Figure 3 shows the effect of x on y1 at temperatures T1, T2,
and T3. Then go on to discuss in detail what is shown on this figure. It may then
be appropriate to conclude with why you expect this property or these results to
relate to property y2, and have the topic sentence to the next paragraph read
The effect of x on y2 is shown in Figure 4. Thus each paragraph covers a single
property or set of results, and the paragraphs flow in a logical order.
As is the case with individual sentences, it is not uncommon to rework and
reorganize a paragraph as you write. Make sure to stick to the topic in the body
of the paragraph. Make your points in a logical order within each sentence,
within each paragraph, and from paragraph to paragraph. Remember that short
paragraphs are more readable than long ones, and are also easier to organize.

59

Some general hints for good technical writing

1 Keep it short! This is true for both sentences and paragraphs. If you find you
are running on too long, try to break up the sentence or paragraph into 2 or more
shorter ones.
2 Stick to the point! The topic of a paragraph should be identified in the topic
sentence. Information unrelated to this topic probably does not belong in this
paragraph. If you find that you need to include additional information in a
paragraph in order to make your points, then you should probably rework the
topic sentence as well.
3 - Make your points in a logical order so the report is easy to follow.
4 Convince your readers! Your presentation should be complete and correct,
so your readers are convinced that you conducted appropriate experiments, that
you conducted them correctly, that your analysis of your data was complete, and
therefore that your conclusions are correct. If you considered alternative
experimental techniques or alternative explanations for your results, present
them and include your reasons for ruling them out. Remember that if you
present a technical paper for publication or at a conference, the reviewers or
audience are likely to try to poke holes in your story, that is, to challenge your
methods or your conclusions. By presenting your complete thought processes,
including the options you ruled out, you can preempt any challenges from your
critics and win their support. You will also have good reason to be confident in
your results!
5 Dont confuse cause and effect. This is a common mistake in student papers
and an important one to avoid! For example, which of the following statements is
correct:
a. The viscosity of the solution increased with increasing concentration of
solute.
or
b. The solute concentration went up with increasing viscosity.
Answer: choice a is correct. Choice b implies that viscosity is the input, or
independent variable (cause) and concentration is the response, or dependent
variable (effect, or output), which is clearly not the case!
6 When discussing data plotted on a figure, be careful not to imply that the x
variable was varied continuously if in fact it wasnt. This is a subtle point but
nevertheless an important one. For example, compare the statement y goes up
with increasing x to y goes up as x is increased. They look almost the same,
but in fact the first example is correct for the case where you measured y at
discrete values of x, and the second example applies when y was recorded over

60

time while x was varied continuously (that is, during real-time monitoring of y with
x changing in real time).
Bad words, or words to avoid or use with caution
1 Be careful with adjectives!
Positive adjectives (such as high, low, large, small, easy, hard, etc.) are
meaningless without numbers or some other basis for comparison (How
high? How low? How large? How small? Easy compared to what? etc.).
They are also meaningless unless you put them in context (a bird is large
compared to an insect, but small compared to a cat).
Comparative adjectives (such as higher, lower, hotter, colder, more, less,
etc.) are meaningless unless you state the basis for comparison (Higher
than what? Lower than what? Hotter than what? Colder than what?
More than what? Less than what? etc.).
Superlative adjectives (such as highest, lowest, largest, smallest, best,
worst) are meaningless unless you define what your sample space is. A
small child may be the tallest person in her class, but is undoubtedly not
the tallest person in the city!
2 Be careful with words that indicate relationships between quantities, such as
increase/increased/increasing, decrease/decreased/decreasing, etc. Here again,
you must state the basis for the comparison (increasing relative to what starting
point?) or the input that is driving the change (y increases with increasing x).
3 Avoid ambiguous verbs such as change, vary, differ, fluctuate, etc., which do
not tell the direction of the effect (does it change up or down?), the type of effect
(does it vary systematically or randomly?), or the specifics of the effect (do they
differ with respect to their response to the input, their color, their odor, some
other property...?).

III. Sample laboratory reports on the refrigerator problem

A complete laboratory report contains information about what you did, why
you did it, what you found, what it means, and how that information can be used.
You may choose to prepare an extensive, detailed report including full sections
on background, methods, results, discussion, conclusions, and references, or a
short report containing a brief summary (sometimes as brief as one sentence!) of
your work. The format you choose depends on who your intended audience is.
For example, a long report is appropriate as an in-house document that may be
referred to by others who plan to work with the same system, or who want to
know everything about what you did and what you found to see how it relates to

61

their own work. It is also the appropriate format for a research paper to be
submitted for publication in a technical journal. A short form of the report, which
contains highlights of your results, is appropriate for presentation to, say, a
manager who wants to know how your work may be used by your company or its
customers, and what its uses are. The reader assumes your experiments were
designed and carried out correctly and that your conclusions are supported by
your data, so its not necessary to provide all the details. However, either way
you must still conduct a complete analysis of your results, as described in
Chapter 8, before you start writing. Below we present long and short versions of
lab reports for the refrigerator problem.

***********************
Complete long version
Investigation of the Response of a CCNY Model 2004 Refrigerator
by A. Student
ABSTRACT
The performance of a CCNY Model 2004 household refrigerator was
analyzed with respect to its equilibrium temperature and time to equilibrium at 3
different positions within the refrigerator, namely, on the top shelf near the light
bulb, in the center, and on the bottom shelf near the back of the unit. The
temperature at each of the 3 positions was measured at hour intervals using
identical thermometers, each immersed in identical, sealed containers containing
equal volumes of water so that the heat transfer properties of the surroundings
were the same for all three thermometers. Results showed that the equilibrium
temperature was the same for all 3 positions at every temperature setting, but the
time to equilibrium was longer by ~1 h near the light bulb than at the other 2
positions, where the equilibration times were the same within experimental error.
The equilibrium temperature inside the refrigerator varies linearly with the setting
on the temperature controller, from 17oC (63oF) at setting 1 to 2oC (36oF) at
setting 6. Linear regression analysis was used to determine the relationship
between setting number and equilibrium temperature.

INTRODUCTION
A refrigerator is a closed, sealed chamber that is maintained at a
temperature below ambient temperature and is used for the purpose of cooling
its contents and keeping them cool. Typical refrigerators consist of a sealed
chamber, a compressor, which removes the heat from inside the chamber and
discharges it into the room, a fan, which circulates the air within the chamber so
as to maintain a uniform temperature throughout, and a light bulb that is switched

62

on when the door to the chamber is opened for the purpose of illuminating the
chambers contents [1]. The actual temperature within the refrigerator depends
on the operating characteristics of the compressor (which are controlled by a dial
accessible from inside the refrigerator) as well as the efficacy of the fan, and is
subject to transient (that is, varying with time) variations due to, e.g., opening and
closing the refrigerator door or placing warm food or containers inside the
refrigerator, as well as spatial variations arising due to heat sources inside the
chamber, such as the light bulb.
This report describes an investigation of the performance of a CCNY
Model 2004 refrigerator. This investigation was motivated by the desire to
calibrate the refrigerator with respect to the equilibrium temperature at each
setting, and also by the observation that at the coldest settings food on the
bottom shelf was sometimes found to freeze, even when the same food placed
elsewhere in the refrigerator did not. Thus in order to establish whether the
temperature is uniform throughout the refrigerator, these experiments were
designed to measure the equilibrium temperature at each setting as a function of
position in the refrigerator, as well as the time to equilibrium on raising the setting
by 1 scale unit.
MATERIALS AND METHODS
The CCNY Model 2004 refrigerator (CCNY Fictitious Appliances, New
York, NY) is a standard 1500 cubic foot household refrigerator with 3 shelves and
a light bulb at the top of the chamber in the back wall, opposite the door. The
temperature is controlled via a dial, accessible from the inside, with settings
ranging from 1 (warmest) to 6 (coolest). The temperature was monitored over
time, at intervals of h, at three positions inside the refrigerator, as follows:
position 1 top shelf next to the light bulb (possible hot spot), position 2
middle shelf near the front, and position 3 bottom shelf near the back of the unit
(suspected cold spot). The temperature was measured using three identical
thermometers (Bronx Standard thermometer Model 1976, Bronx Instrument Co.,
Bronx, NY). The thermometers were immersed in equal volumes of water
contained in three identical 150 mL PyrexR beakers sealed with ParafilmR to
prevent evaporation. Thus the heat transfer characteristics in the vicinity of all
three thermometers were the same. Time intervals between temperature
readings were measured using a stopwatch.
The system was initialized by turning off the refrigerator, emptying it
except for the 3 beakers of water, and allowing it to reach ambient temperature.
Then the unit was turned on to setting 1 and the temperature was measured at
each of the 3 positions at intervals of hour until 3 successive readings showed
a constant temperature. The setting was then increased by 1 scale unit and the
procedure was repeated, continuing through to the highest scale setting. Care
was taken to make the measurements as rapidly as possible so as not to leave

63

the door open any more than necessary. Note that the refrigerator was empty
throughout the experiment except for the 3 beakers of water.
RESULTS AND DISCUSSION
Typical results showing how temperature varies with time at all 3 positions
are shown in Figure 1, using setting 4 as a representative example. The
temperature starts out at the equilibrium value, which is necessarily the same for
each position. Then the temperature drops off with time until it reaches its new
equilibrium value. However, the rate at which the temperature drops off is
slowest at position 1 (near the light bulb), but equal for the other two positions. In
addition, the time to equilibrium is longest for position 1 as well.
Figures 2 shows the equilibrium temperature corresponding to each of the
6 settings on the temperature controller. The equilibrium temperature decreases
linearly with increasing setting number, ranging from 17oC (63oF) at setting 1 to
2oC (36oF) at setting 6. Thus Figure 2 may be used as a calibration curve for the
CCNY Model 2004 refrigerator. Linear regression analysis of the data on this
figure gives the following functions:
setting = (20 - t(oC)) / (3)

(1)

or
setting = (68.4 - t(oF)) / (5.4)

(2)

for temperature in Celsius or Fahrenheit units, respectively. These relationships

may be used calculate the correct setting for any required temperature between
17oC (63oF) and 2oC (36oF).
Figure 3 shows the time to equilibrium for each of the 3 positions at each
setting. Here the time to equilibrium is defined as the time corresponding to the
first measurement giving a temperature that was within 0.5oC of the final
equilibrium value. Better resolution could have been obtained if the
measurements had been made over shorter time intervals; however even with
this approximation the data clearly show the existence of a hot spot near the
light bulb, where the time to equilibrium is longer than elsewhere inside the
refrigerator. This may be attributed to the heat generated by the bulb itself. The
fact that the temperature at position 1 ultimately does reach the same value as
the rest of the chamber indicates that the fan is effective in circulating the air
throughout the chamber. The time to equilibrium is 3.0h at position 1 and 2.0h at
positions 2 and 3, and is the same for all settings within the resolution of these
data. Note that we did not observe any evidence of a cold spot at the bottom of
the unit, contrary to our earlier observations. It may be that this effect only
occurs when the refrigerator is full and the contents interfere with the circulation
of the air. Additional experiments are required to test this hypothesis.

64

CONCLUSIONS
The CCNY Model 2004 refrigerator is effective in cooling its contents over
the range 17oC (63oF) to 2oC (36oF), with a linear relationship observed between
the equilibrium temperature inside the unit and the setting on the temperature
controller. There is a transient hot spot in the vicinity of the light bulb due to
heat generation by the bulb; however even at this position the temperature does
reach equilibrium after 3.0h under the conditions of this experiment, compared
with 2.0h to equilibrium elsewhere in the unit. There was no evidence of a cold
spot in the empty refrigerator, although this problem has been observed when
the unit is full.
REFERENCES
1. Black, A.B. and White, C.D. Fictitious Textbook on the Design of Heating and
Refrigeration Units, 3rd ed. McGraw-Hill (New York) 2001 pp. 307-315
**********************
Sample short versions
1. The Abstract of the long version presented above is an example of a short
version of a complete report. It contains all the important information about your
work, with some practical quantitative information supplied as well.
2. The CCNY Model 2004 refrigerator is characterized by a hot spot in the
vicinity of the light bulb, where the temperature may be elevated
temporarily relative to the rest of the interior of the refrigerator. This onesentence report tells your conclusion about one aspect of the operation of the
refrigerator. It would be appropriate as, say, a weekly highlight, presenting
something unusual, unexpected or of particular interest that you learned from
your experiments. It is also appropriate for an audience that has a specific
interest in knowing whether the temperature is uniform throughout the
refrigerator, or if there is anything unexpected in the way this refrigerator
behaves.
3. Based on our studies on the temperature profiles in the CCNY Model
2004 refrigerator, in which we demonstrated that the time to equilibrium
temperature throughout the refrigerator is 3.0h after changing the
temperature setting, we would not recommend this model for the
application of interest to Client 12. This is a report with an impact! You are
making a recommendation based on sound engineering practice that led to
quantitative conclusions and an assessment of the clients needs.

65

4. Find out what question your reader wants answered, and present the answer
with as much data or discussion to back it up as is required to convince your
reader that your response is correct.

66

Chapter 7: UNIT OPERATIONS LABORATORY HARDWARE

Unit operations in chemical engineering are processes involving the
physical (as opposed to chemical) treatment of solids, liquids, or gases, including
mixing, separating, pumping, heating, cooling, evaporating, condensing, drying,
humidification, etc. A unit is a device or vessel in which one of these processes
is carried out. Typically, two or more units are connected together in an
integrated process that involves several of these operations, requiring the use of
hardware such as pipes, tubing, junctions, valves, meters, traps, containers, etc.
It is important to be familiar with the many different types of hardware and
devices that are available, and with the specific limitations regarding the use of
each type.
The best way to start to learn about the variety of hardware options
available is to read through catalogs of general laboratory supplies, some of
which are listed in the Bibliography. You may also access many of these
catalogs on line. You will find that for every category of equipment and hardware
there are many different products, each with different specifications regarding
safe operating ranges of pressure and temperature, as well as chemical
resistance. As an example, at the end of this chapter there is a copy of a page
from a Cole-Parmer Industrial Catalog 2004/05, page 357, showing different
types of flexible tubing. You will note that the products on this page include
tubing made from 2 different materials (peroxide-cured silicone and platinumcured silicone), both reinforced and not reinforced, with maximum pressure
ratings at 70oF ranging from 8 psi to 312 psi. Now note that there are 9 pages
just for flexible tubing listed in the index of this catalog! Thus it should almost
always be possible to meet whatever specific hardware needs you have under
ordinary operating conditions. But dont assume that an existing system
necessarily meets your requirements! One semester in ChE 460 there was a
loud BANG when a rubber hose running directly out of the compressed air
receptacle in the wall exploded! Fortunately no one was hurt, but the group
working on that experiment found that the hose was rated for pressures only up
to 10 psi, while the pressure in the line was of order 80 psi! What to do? The
group designed a pressure reduction system consisting of a series of step-down
valves connected by appropriately-rated rubber hose, to protect the apparatus
that was using the compressed air as well as the safety of the people in the
laboratory!
Below is a very abbreviated list of measuring and control devices
commonly used in chemical plants and laboratories. It is left up to the student to
learn how the different devices work, how they are used, and what the
differences are.

67

Measuring devices
Temperature
thermometers
thermocouples
Pressure
manometers
pressure gauges
Flow rate
rotameters
paddle-wheel flowmeters
Doppler flowmeters
insertion flowmeters
etc.
Controllers for each device, find out what it senses and how it responds!
Temperature
various types of electronic temperature controllers
temperature switches
Pressure
valves needle valves, ball valves, check valves, relief valves, etc.
regulators
Flow rate
various types of flow controllers; check different catalogs to read about
patented designs!

68

BIBLIOGRAPHY
General reference Texts
Perry, R.H. and Chilton, C.H. Chemical Engineers Handbook, any recent edition, McGraw-Hill
(New York)
Dean, J.A., ed. Langes Handbook of Chemistry, any recent edition, McGraw-Hill (New York)
rd

Shoemaker, D.P., Garland, C.W., and Steinfeld, J.I. Experiments in Phbysical Chemistry, 3 ed.
(or latest edition) McGraw-Hill (New York) 1974
Unit Operations Texts
Levenspiel, O. Engineering Flow and Heat Exchange Plenum Press (New York) 1998 or latest
edition
th

McCabe, W.L., Smith, J.C. and Harriott, P. Unit Operations of Chemical Engineering, 6 ed. or
any other recent edition, McGraw-Hill (New York) 2001
Treybal, R. Mass Transfer Operations, 2

nd

ed. McGraw-Hill (New York) 1980?

op. cit. Perry, et. al.

Statistics texts
Levine, D.M., Ramsey, P.P., and Smidt, R.K. Applied Statistics for Engineers and Scientists
R
R
Using Mikcrosoft Excel and MINITAB Prentice Hall (Upper Saddle River, NJ) 2001
op. cit. Shoemaker, et. al.
op. cit. Perry, et. al.
Safety texts
Crowl, D.A. and Louvar, J.F. Chemical Process Safety: Fundamentals with Applications Prentice
Hall (Englewood Cliffs, NJ) 1990 or latest edition
Texts on good writing
nd

Strunk, W. Jr. and White, E.B. The Elements of Style, 2 ed. The Macmillan Co. (New York)
1972 (a wonderful and very readable (and short!) guide to usage and style. Available in
paperback.)
Platt, H. Jr., Guide to Usage, in The American College Dictionary Random House (New York)
1968 p. 1425 (probably included in more recent editions as well. Also, look for similar
articles in other dictionaries).

69

Suppliers and catalogs

Cole-Parmer, Vernon Hills, IL (www.coleparmer.com)
1-800-323-4340
General Catalog
Industrial Catalog
Fluid Handling Systems Sourcebook
Instruments for Environmental Health and Safety
R
R
Encyclopedia Masterflex buyers guide for Masterflex tubing pump systems
FoodTech Source for produces for food and beverage processing and testing
Tubing, Fittings, Valves, & Flowmeters catalog
Fisher Scientific, (www.fishersci.com)
1-800-766-7000
Extensive catalog of general laboratory supplies and some Fisher brand analytical
instruments.
VWR, (www.vwrsp.com)
1-800-932-5000
Extensive catalog of general laboratory supplies and some VWR brand analytical
instruments
Thomas Scientific, Swedesboro, NJ (www.thomassci.com)
1-800-345-2100
Extensive catalog of general laboratory supplies