~~~~~~-~~~~~~

MINERAL ENGINEERING

Development of a model for predicting the

screening performance of a vibrating screen
V.K. KARRA
Senior Research Engineer
Process Machinery Division, Rexnord
Milwaukee, U.S.A.

ABSTRACT
A mathematical model for the performance of a vibrating
screen is developed in terms of an oversize partition curve.
This is a normalized curve for any assemblage of particles
presented for screening when their size is normalized with
respect to the 50 per cent passing size (d50 ) under the prevail-
ing screening conditions. Based on the throughfall aperture
(hT) of the wire mesh on the deck, the effects of oversize
(% > h T), half-size under (% < 0.5 h T), deck location and feed
rate were evaluated using the existing screen sizing informa-
tion. All these variablesand also the near-mesh (1.25 hTx O. 75
h T) content (%) of the feed affect the d50 •
Performance data from different decks of a double-deck
screen with various feed rates, size distributions and screen
apertures were used to evaluate the near-mesh factor. Effects
of wet screening and material bulk density are also incor-
porated in the model. This generalized model is useful for the
simulation of a multiple-deck screen. It furnishes the sieve
,.
FIGURE 1. Photograph of the test screen system.
analysis information of all the deck output streams and
therefore has a potential use in the over-all circuit evaluation
when used in conjunction with relevant crushing and grinding
performance models. An example illustrating the prediction tions, very often a vibrating screen is employed. To be able to
capabilities of the model is given. predict how a screen is going to perform under different feed
conditions, the development of a screening process model is
Introduction necessary. The model will be of great use both in effici ent
Screening is a widely used industrial process for the continuous screen sizing and in simulatio n of closed-circuit crushing. For
separation of materials, based on particle size. For processing the latter purpose, Whiten'D and Mular(2) have developed
coarse particle assemblies, prevalent in the crushing opera- screen models. Both these workers did not use con ventional
screen sizing methodology in their model development.
With Rexnord's role as equipment manufacturer, we were
interested in developing a performance prediction model based
on the existing screen sizing methodology. The idea, of course,
V. K. Karra is to relate it backwards for the development of an efficiency-
based screen sizing method . All efforts were made to keep the
Vijia K. Karra is now a manager, pro- developmental process simple enough for an easy understand-
cess research, after being a senior ing by various levels of screen application people.
research engineer for over 2 years in
the Process Machinery Division of Experimental
Rexnord, Milwaukee, U.S.A. He During 1973, data were gathered from a Nordberg double-
holds a Bachelor's degree in deck 1.52- by 3.66-m GP screen in the Waukesha Lime and
metallurgical engineering from India Stone plant in Wisconsin. It is a two-bearing circle-throw in-
and Master and Doctoral degress in clined screen (20 degrees to horizontal), with an SOO-rpm
materials science (minerals) engineering from the University of motor and 7.9-mm throw. The screen system (Fig. I) was set
California, Berkeley. Prior to his graduate study, he was a senior up for convenient sampling and tonnage mea surements on
scientificassistant for over four years in the Minerals Division of the various streams. Variables tested were as follows :
Regional Research Laboratory, Bhubaneswar (India). His current in- Feed Rate (tph):103 to 299
terests are in the processanalysis and modelling of crushing, grinding Feed Size (mm): 50.8xO, 38.lxO, 25.4xO, 18.8xO
and screening equipment. Mesh on Top Deck (mm sq.): 25.4, 22.2, 15.9
Mesh on Bottom Deck (mm sq.): 15.9, 9.53, 5.41
Keywords: Mineral engineering, Models, Screening, Vibrating
Data on tonnage and sieve analysis of feed, top-deck oversize,
screens, Partition curves, Sieve analysis, Metallurgy.
bottom-deck oversize and bottom-deck undersize were compil-
ed.

elM Bulletin, April 1979 167

Process Analysis basic shape of the curve will be as shown in Figure 3.
Mathematical formulation of the curve would be very useful in
Figure 2 is a schematic showing the input (feed) and outputs
closed-circuit crushing simulations . Usually, such a formula-
(oversize and undersize) of a screen deck. Let QF, Qo and Qu tion consists of a parameter, d so, the size corr esponding to the
be the tonnages and f F, fo and fu be the mass fractions in a par- 50 per cent partition value (of undersize or overs ize). Process
ticular size interval for feed , oversize and undersize streams parameters, such as the -oversize, half-size under, near-size
respectively. fractions , deck location, feed rate and bulk density, affect d so•
Under steady-state screening conditions , the over-all mass
Depending on d so (Fig. 3), the curve may be shifted to the left
balance provides- or right, giving poorer or better screening efficiency for that
QF = Qu + Qo· · · · · · · ·· ······ ·· ··· · ·· · · · ··· ·· ·· · (I) deck. It is convenient to normalize such curves by plotting C,
The mass balance of material in any size interval, i, gives- versus di/d so for all experimental condition s and screen decks.
QFfF(i) = Qofo(i) + Qufu(i) (2) As one can visualize, all such curves pass through the point
Measurement of screen performance through partition C, = 50010 and d;/d so = I. An average curve, considering
curves, based on the probability distribution of the passage of the scatter in the top and bottom portions, can be taken as the
particles with respect to size, was suggested by several normalized partition curve. Once d so is known, this curve or its
workers(3,4,S). Hydrocyclone classifiers have been relatively equationcan be used for figuring out the partition coefficients
well characterized in this respect(6.7). The importance of for all the size fractions of the feed to the deck.
classifier characterization was extensively discussed by Austin Based on our objective to simultaneously accomplish an im-
and Luckiew. To represent a non-ideal classification operation proved screen sizing method, we correlated screen efficiency
like the actual screening process, we use oversize partition (for undersize recovery) to d so• The efficiency in turn was
coefficients for different size intervals. The ratio of the weight related to the various screen factors given in the Nordberg Pro-
of material in the ith size interval of the oversize stream to the cess Machinery Reference Manual'?'. Although all the factors
weight of material in the same size interval of the feed stream is that affect the screening performance are qu antified in the
represented as the oversize partition coefficient, C;, for the manual, the near-mesh factor is not given . A feed consisting of
material in that size interval. a large fra ction of particles of a size closer to the screen aper-
C, = ~ . . . . . .. . . . . .. .. . • . . ....... . . . . . . . . . . . " . . (3) ture would present a more difficult separation than a feed con-
QFfF(i) sisting of a very small fraction of such near-mesh particles.
(I -C0 is the undersize partition coefficient. The maximum Therefore, an assessment of the near-mesh factor was also re-
and minimum values of these coefficients are unity and zero quired.
respectively. When the oversize partition values, C;, are plot- The terms oversize, near-size, etc., are referred to with
ted against the corresponding geometric mean particle size, respect to the throughfall aperture of the screen mesh on the
d, [= J(d 1d2)d , of the material in the ith size interval, the deck. For a deck inclined at an angle (J to the horizontal with a
square mesh of aperture h and wire diameter d, the throughfall
aperture, hT , is given by:
hT = (h+d) cos () - d (4)
Following the Vibrating Screen Manufacturers Association(lO)
terms and definitions:
0/0 oversize is 0J0 > hT
% half-size und er is % < O.5hT
% near -mesh is % in the size interval 1.25 hTxO.75hT
Screen sizing is generally done either on a feed tonnage basis
or on a theoretical undersize tonnage basis. Our practice is to
go by the latter. The sizing is done using the relation:
Screening area = Theoretical undersize tonnage (tph) .. . . (5)
UNDERSIZE STREAM required (sq.m) ABCDEF
Qu.fu
where A,B,C,D,E,F are respectively basic capacity, oversize,
half-size, deck location, wet screening and material weight fac-
tors. Because A has the tph/sq.m units, both sides of equation
5 are dimensionally consistent. A dimensionless group,
FIGURE 2. A scbematic sbowing tbe input (feed) and outputs
K = Theoretical undersize tph I sa.m of screening area (6)
(oversize and undersize) of a screen deck .
ABCDEF
can be formulated .
Performance of a screen deck can be measured by the effi-
ciency of undersize recovery defined as:
100 efficiency = 'I. of feed (or amount) which actually passes through (7)
'7. of feed (or amount) which is undersize (should pass through)

80
TABLE 1.
% Cj
r 60
50% Size Interval dl 'F ,1
0
,2
0
,2u ,1u c] c~
(d-x d21 mm (=W21I) 147 lph 27.2 32.7 87.1 119.8 % %
40
37.6x26.7 31.7 .0167 .09 99.72
20 26.7x18.8 22.4 .1129 .55 .05 .0136 90.14 100.0
18.8x15.9 17.3 .1497 .28 .44 .1201 34.61 100.0
15.9x12.7 14.2 .0975 .05 .37 .01 .1083 9.49 93.25
0- L...-_..JIC..._ _--+---':....::...._ _ 12.7x9.53 11.0 .1130 .03 .11 .14 .1318 4.91 22.78
9.53x7.94 8.70 .0600 .03 .09 .0736 11.13
d _ 7.94x6.68 7.28 .0652 .11 .0800
i 6.68x4 .70 5.60 .0948 .16 .1163
4.70x3 .33 3.96 .0830 .14 .1018
(= J (d, d 2 ) j ) 3.33x2 .36 2.80 .0592 .10 .0727
< 2.36 .1482 .25 .1818
FIGURE 3. Basic sbape of an oversize partition curve for a
screen deck. "Superscripts 1, 2 refer to top and bottom decks respectively.
• ·dJo = 18.71 mm, ~ = 12.24 mm

168
 \
) ·O-r--------------~
100

U (()~
e, 80 0·8
ill tl
tJ;
o
o
N ....
'ij)
'-
ill
60
s
~ o
o

> r)'S >. 0·6 o

0 o
....0
(>;)\;; c
tn
~ ill

....g
C
c
.;:; 40 ..:. 0·4
'-
0 R
(lJ
ill
Q. ~
ill
'-
.... 20 ~
C
~
II
0·2
ill
o'- a'
0\'
ill
Q.
0- O-,l'----+----+---+------i---,
0'4 06 0·8 1·0 1·2 1-4
o 0·2 0·4 0·6 0·8 )·'0
(dj/d s O)
relative aperture (dSO/h t )
FIGURE 4. Normalized oversize partition curve.
FIGURE 5. Tbe Interdependence or relative aperture and
screening efficiency: top deck.
Using a multiple regression computer program, this efficiency
can be correlated to K and the percentage of near-mesh
material in the feed, X n , as:
)·0-.-------- - - -- - -- - ---"'7\
efficiency = a (K)b (1- ~)C . . .. . • .. . • . . . . . . .. . . . . . .. . .. (8)
100
Equation 8 permits the development of a near-mesh factor G,
defined as: 0·8
G = a ' (1- ~)c· (9)
100

Results and Discussion >. 0·6

o
Table I shows the results of oversize partition value calcula- c
ill
tions from the data gathered on the 1.52- by 3.66-m double- .~
deck GP screen, having a 22.2-mm-sq. mesh on the top deck ....ill
.... 0·4
and a 15.9-mm-sq. mesh on the bottom deck. Feed rate was
147 tph. Corresponding calculated d so values are also given. In
a similar manner, calculations were done for the rest Of the.19 02
tests. The plots of C, versus dj/d so for each test aperture and
feed rate for both top and bottom decks were drawn and
superimposd one over another and the normalized partition
curve was constructed (Fig. 4). This curve can be represented O-,l'---+----+----+----+---~
by the relation: o 0·2 0·4 0·6 0·8
0J0 Cj = 100 [1- exp(-0.693(d j/d so)5.846») ......••••.•.. (10) relative aperture (dSO/h t )
The moisture content of the feed materials used in these
studies was below 5%. Moistures above this level could create FIGURE 6. Tbe Interdependence or relative aperture and
screening efficiency: bottom deck.
conditions for by-pass of fines to the oversize stream. This
phenomenon and variables such as screen throw and speed
could influence the shape of the normalized partition curve .
Particle shape could be another significant variable, calling for are differentw. In the present study, involving dry screening of
testing with various materials . In the future, attempts will be crushed stone, factors E and F are both equal to unity .
made to assess the engineering significance of these variables . For each set of experimental data factors A,B,C,D,
In Figures 5 and 6, efficiency is plotted against the relative theoretical undersize tph/sq.m, per cent near-mesh fraction in
aperture, dso/hT , for top and bottom decks respectively. The the deck feed and efficiency were calculated, and then multiple
straight lines in these figures correspond to the relation: regression analysis for equation 8 gave the values a = 0.975,
efficiency = d soI h T (II) b = -0.148 and c = 0.511. From this relation, equation 9
It is clear that the relative aperture is directly controlling the for the near-mesh factor G was developed. The values of a '
screening efficiency. A physical interpretation of this observa- and c ' are 0.844 and 3.453 respectively.
tion would be interesting and perhaps may provide an alter- In the Appendix, a worked example for the calculation of
native definition for screening efficiency. Again, this relation- tonnage and sieve analysis for the undersize and oversize from
ship between relative aperture and screening efficiency could top and bottom decks is given. In this example, the top and
be influenced by some of the variables mentioned in the con- bottom decks had 15.9-mm-sq.-mesh and 5.4I-mm-sq.-mesh
text of the normalized partition curve. medium-gauge wire clothes on them. Minus -18.8-mm stone at
Equations 12 to 31 for factors A through F are given in a tonnage of 103 tph was the feed to the top deck. Top-deck
Table 2. Equations for factor A are for crushed-stone applica- calculations are done first. Throughfall aperture is calculated
tion. Generally, the same A values are used for sizing screens by equation 4. Usually , the hT value falls within the upper and
in metallic ore applications. For sand and gravel, these values lower size bounds of an interval. Calculations of percentages

elM Bulletin, April 1979 169

 100- r--------------------------""7't"::.&---=-€~-____,-O

0 90
_ _ CALCULATED DISTRIBUTION 10
Z o A ACTUAL DISTRIBUTION
'" 80
o
20
'"
~
TOP DECK
w 70 A BOTTOM DECK 30
>
....
«..... 60 40

::l 50 50
:E
::l
U 40 60

~
..
.... 30
TONNAGE
TOP DECK BOTTOM DECK 70
::E: (TPH) I 2 3 4
0 20 OVERSIZE UNDERSIZE OVERSIZE UNDERSIZE 80
w 21 82 49 33
OBSERVED
~ 10 90
CALCULATED 20 83 49 34-
0 -L.---- -+~:......+____4~'_I~I___1I__I_!.-IO(

-0 ~ R ~ lri r-, <Xl r-, -0 <Xl

M .... o-:N<Xl'O"'O
C'l M""f.... C'lMll')
PARTICLE SIZE (mm)
FIGURE 7. Comparison of actual and calculated tonnages and sieve analyses for a 1.52- by 3.66-mm DDGP screen with 15.9- and
5.41-mm-sq.-mesb clotbs , respectively, on tbe top and boltom decks. Material: -18.8-mm crushed stone. Feed rate: 103 tph.

of oversize, half-size and near-size material in the feed are size and undersize and sieve analysis (see Table 3) are fairly
done using appropriate linear interpolations. For example: straightforward. A similar calculation procedure gives the per-
Q (>14.6) = 6.32 + 14 (15.9-14.6)/(15.9-12.7) = 12.01% formance information for the bottom deck (Table 4). The
R «7 .30) = 7.11 (7.30-6.68)/(7.94-6.68) + 10.74 + over-all information gathered from this simulation procedure
7.42 + 5.68 + 21.47 = 48.81 % is given in Figure 7. Comparison of the experimental and
X, (18.3 x 11.0) = 6.32 (18.3-15.9) / (18.8-15.9) + 14 + calculated values shows fairly good model predictions. Figure
20.42 (12.7-11.0) / (12.7-9.53) = 30.18070 8 shows the performance summary for the 1.52- by 3.66-m
Here, d 50 is calculated from double-deck GP screen with 25.4-mm-sq.-mesh on the top
d50 = hT ( Theo. undersize t Ph/s Q.m)-O.l411 (12) deck and 15.9-mm-sq.-mesh on the bottom deck . Minus
ABCDEFG 50.8-mm crushed stone was fed at 207 tph to the top deck . The
Oversize partition values are calculate using the d 50 in equa- simulation capability of the model is once again fairly good.
tion 10. Once this is done, the rest of the calculations for over-
Conclusions
TABLE 2. Equations for various screen factors The vibrating screen performance model developed by coup-
given in the Reference Manual(9) ling the existing screen sizing methodology with a normalized
oversize partition curve gave fairly good predictions of the
tonnages and sieve analyses of the oversize and undersize pro-
Factor A (for crushed stone) P - opening s ize (mm) = hT
A = 12.1286 (P)O.3162.-10.2991 P<50.8 (12) ducts of a double-deck GP screen processing crushed stone .
A 0.3388 (P) + 14.4122 P~50.8 (13) The model is expected to be of potential value in closed-circuit
Factor B 0 = % oversize in feed to deck crusher simulation work and in the development of an improv-
B -0.012 (0) + 1.6 Os87 (14) ed screen sizing methodology.
B 0.0425 (0) + 4.275 0>87 (15)
Factor C R % half-s ize In feed to deck
C 0.012 (Rl + 0.7 Rs30 (16) TABLE 3.
C = 0.1528 (R)O.564 30<R<55 (17l
C = 0.0061 (R)1 .37 55sR<80 (18
C = 0.05 (R) - 1.5 R~80 (19) TOP DECK
Factor D S = Deck location; top: S =1, Second S =2, etc. Size Feed 0/0
D = 1.1 - 0.1 (S) (20) (mm) (tph) Reporting Oversize Undersize
Factor E for wet screen ing, let T = 1.26 hT to (tph) (tph) (%)
E = 1.0 T<1 (21) Oversize
E T 1 sTs2 (22)
E = 1.5 + 0.25T 2<T <4 (23) 18.8x15.9 6.51 95.86 6.24 0.27 0.33
E = 2.5 4 sT s6 (24) 15.9x12.7 14.42 63.63 9.18 5.24 6.33
E = 3.25 - 0.125T 6<Ts10 (25) 12.7x9.53 21.03 20.27 4.26 16.77 20.27
E 4.5 - 0.25T 10<T < 12 (26) 9.53x7.94 7.05 5.58 0.39 6.66 8.05
E = 2.1 - 0.05T 12sTs16 (27) 7.94x6.68 7.33 2.01 0.15 7.18 8.68
E 1.5 - 0.125T 16<T<24 (28) 6.68x4.70 11.06 0.44 0.05 11.01 13.31
E = 1.35 - 0.00625T 24 s Ts 32 (29) 4.70x3.33 7.64 0.06 a 7.64 9.23
E 1.15 T>32 (30) 3.33x2.36 5.85 0.01 0 5.85 7.07
Factor F U = bulk density (kg/m 3) <2.36 22.11 a 0 22.11 26.73
F =
(U/1602) (31) TOTAL 103.00 20.27 82.73 100.00

170
 100- 0

e 90 CALCULATED DISTRIBUTION 10 C
u.I
Z o l:i. ACTUAL DISTRIBUTION Z
Vi 80
V)
o TOP DECK
20 «
~
~
u.I
u.I 70 c: BOTTOM DECK 30 at:
> u.I
>
<
.....
::>
60 40
~
«
.....
50 50 ::>
~
::> s:::J
u 40 60
U
~

~
~
30 TOP DECK BOTTOM DECK 70 &i
TONNAGE ~
J: ( TPH) I 2 3 4 J:
e 20 OVERSIZE UNDERSIZE OVE RSIZE UNDERSIZE 80 o
u.I
OBSERVED 103 104 33 71 u.I
~ 10 90 ~
CALCULATED 98 109 37 72
0 .-100
co /'0.. co /'0.. 00 co
00
('")
('")
('")
0
/'0.. 00 &i r...:
N co .0 ('") 0
N -.i -o 0:
PARTICLE SIZE (mm) ('")
N l£)

FIGURE 8. Comparison of actual and calculated tonnages and sieve analyses for a 1.52- by 3.66-mm DDGP screen witb 25.4- and
15.9-mm-sq.-mesh cloths, respectively, on tbe top and bottom decks. Material: -50.8-mm crushed stone. Feed rate: 200 tph.

Acknowledgments 17, Mineral Processing Plant Design; Editors: A . L. Mular, R. B.

Bhappu; Soc. of Mining Engineers, A/ME. New York, 1978.
The author sincerely acknowledges James R. Gronseth, vice- 8. LUCKIE, P. T ., and AUSTIN, L. G., Technique for Derivation
president, engineering, at Rexnord for permission to publish of Selectivity Functions from Experimental Data, 10th Int.
this paper and William J. Hetzel, product manager, screens, Mineral Proc. Congress, London. /973. pp. 773-790.
for the provi sion of his painstakingly compil ed field data . 9. Nordberg Process Machinery: Reference Manual, Rexnord, Ist
edition, /976. pp. 14-8. 14-9.
REFERENCES 10. Vibrating Screens: Terms and Definitions, Vibrating Screen
I. WALTER, G. W., and WHITEN, J . W., An Examination of Manufacturers Association. New York, /967.
Tertiary Screening Using Simulations, A us. / .M.M., No. 26/. Appendix EXAMP LE
March /977. pp. 13-/6. Feed 10 3 rpb
2. MULAR, A. L., and HERBST, J. A ., Digital Simulation: 1. 52 x 3. 66m DOCP Sc r e e n Bot t ca de ck : 5. 41 mm..sq. Mesh
Top deck: 15.9 mm. . sq. Mes h 3 . 25mm Wire 01a .
An Aid for Mineral Processing Plant Design, Chap. 14, Mineral 4. 8Bmm Wir e 01 a .
Processing Plant Design; Editors: A. L. Mular, R. B. Bhappu; Feed Analysis
~~ Perc en tage
Soc. of Mining Engineers. A/ME. New York, /978.
Siz e (mill) Percentage 7. 94x6.68 7.11
3. BRERETON, T ., Performance Measurement of a Screening
Process Using Statistical Techniques, Quarry Managers J.• Vol. 18 . 8xI 5. 9 6 . 32 6.6 8x4 . 70 10 . 74
54. Sept. /970, pp. 344-346. 15.9x12.7 14 . 00 4.70x 3 . 3 3 7 . 42
4. LAVERICK, M. K., Size Classifier Performance, Mine &
12 .7 x 9.5 3 20 .42 3 . 33x2 . 36 5.68
Quarry, Sept. /975, pp . 33. 35, 38. 43, 46.
5. PATRIDGE, A. C ., and ROBERTS, J ., Principles of Screening, 9 .5 3x7. 94 6 .84 <2.3 6 21.47
Mine & Quarry, December /977, pp. 33-38. To simulate the p erforman ce o f the sc r e e n wit h re spect t o tonn a ge a nd
6. LYNCH, A. J., and RAO, T . C., Modeling and Scale-Up of
s ize dis tribution anal yse s o f the oversize an d und e r s ize s treams f r cm
Hydrocyclone Classifiers, 11th Int . Mineral Proc. Congress,
Cagliari, /975. the t op and bottom decks of t h e s creen.
7. MULAR, A. L., and JULL, N. A. , The Selection of Cyclone Top Deck Calculations
Classifiers, Pumps and Pump Boxes for Grinding Circuits, Chap.
hr· 14 .6 ... . Q - 12.01% , R - 4 8 . 8 1%, X - 30 . 18%
D

TABLE 4. A = 18.01. B = 1.4559 , C = 1.3691, 0 = I, G = 0. 2441

BOTTOM DECK Theoretical undersize ( tph/aq:m) = 16.29

0/0
dSO = 13.32.0
Size Feed Reporting Oversize Undersize
(mm) (tph) to (tph) (tph) %C1 - 100 {I - exp ( - 0 . 6 93 (dV lJ . 45 ) 5.846 )]
Oversize Th e und e r s t ae from the to p de c k i s th e f e ed to t h e b o tt o m deck .

18.8x15.9 0.27 100 0.27 0 BOtt OCl Deck Calcula tio ns

15.9x12.7 5.24 100 5.24 0 hr - '. 89"". • Q (>4. 89 ) ·43.66 + 13.31 ( 1.;;/ ' . 9 8) . 55.69%
12.7x 9.53 16.77 100 16.77 0
9.53x7.94 6.66 100 6.66 0 R « 2 . [.5 ) · 26 .1 3 T 1 . 0 7 ~0.09/0 .97) - 27 .39%
7.94x6.68 7.18 100 7.18 0 x" ( 6 . 11 x 3 .6 7) • 13. 31 <1. 41/1.9 8) + 9 .23 ( 1.03/1. 37) • 1 6 . 42%
6.68x4.70 11.01 91.50 10.07 0.94
4.70x3.33 7.64 27.54 2.10 5.54 A ' 9.7 4. B · 0.9317 , C. 1.0263, 0, 0 . 9 . G ' 0 .4 54 3
3.33x2.36 5.85 4.21 0.25 5.60 The oret i c al un de rs i ze ( t p h / s q .m . ) . 6. 59
< 2.36 22.11 0 0 22.11
dSO • 4. 5lmm
TOTAL 82.73 48.54 34.19
% C, - 100 11 - e xp (-0 .693 (dI/4. S1 )5. 846 ) J

elM Bulletin, April 1979 171