0 views

Uploaded by Anjneya Srivastava

sefee

- MODELING AND OPTIMIZATION OF A MOBILE CROP RESIDUE DISINTEGRATOR
- STATISTICS IN RESEARCH
- Basic Formulae for Mixed Effects Models in R
- ac.els-cdn.com_S092401360200273X_1-s2.0-S092401360200273X-main
- FELIX RESULT.DOC
- Reducing Variability With Experimental Design
- BF1
- Research Plan Tahong
- l13_ancova
- Statistical Analysis and Optimization of Ammonia Removal From Landfill Leachate by Sequential Microwave_aeration Process Using Factorial Design and Response Surface Methodology
- 12
- 21 Inter Cropping
- Alternatives to Difference Scores_Polynomial Regression and Response Surface Methodology_.pptx
- Tute Exercise 11 Solutions1
- 15
- LAB Tortoise Hare
- Analysis of TRAI Decay
- Example of Lab Report Mark Scheme
- 79-86
- 3169004_1073605809_Labreportguidelines-2018-2019

You are on page 1of 23

Source: Supplement to the Journal of the Royal Statistical Society, Vol. 6, No. 1 (1939), pp.

1-22

Published by: Wiley for the Royal Statistical Society

Stable URL: https://www.jstor.org/stable/2983620

Accessed: 30-11-2018 06:34 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

https://about.jstor.org/terms

Royal Statistical Society, Wiley are collaborating with JSTOR to digitize, preserve and

extend access to Supplement to the Journal of the Royal Statistical Society

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

SUPPLEMENT

TO THE

AND AGRICULTURAL RESEARCH SECTION OF THE ROYAL STATISTICAL

SOCIETY, November 24th, 1938. PROFESSOR R. A. FISHER, F.R.S., in

the Chair.]

discussion, that I regard myself as an expert on animal experimenta-

tion, or that I have necessarily a great deal of experience of the

subject. The statistician must, as always, tread warily in dealing

with problems which are the acknowledged sphere of the practical

expert, but his advice on lay-out and statistical interpretation of

results may be useful, advice which should not be given without a

study of the practical problems involved. Nor could I have hoped,

in the very short time which has been available since the discussion

was planned, to study afresh the extensive literature on the subject,

in order to deal comprehensively with all animals, or with all experi-

mental problems. My purpose will be served if I am able to set the

ball of the discussion rolling.

My first contact with experiments of this nature was when, some

ten years ago, I examined the figures of certain trials laid out to test

the effect of treating grass-land with different grades of basic slag.

In some the effect was measured by the weight of hay taken off the

land, in others by the live-weight increase of animals fed on the

pasture. I recall one experiment where three, or at most four,

plots IO acres in size, each having a different treatment from the

others, were grazed by cows. Weekly weights of individual animals

were recorded, and the differences between initial and final weights

over the same time interval were averaged for each treatment, and

used as a measure of the effectiveness of that treatment. In a sense

this is not an animal but a crop experiment, but because the animal

was introduced to measure the effect, there were introduced at

the same time the difficulties which attend animal experimentation.

Chief of these was the large error involved, but there were others.

Practical considerations required the effective grazing down of the

SUPP. VOL. VI. NO. 1 B

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

2 WISHART-Statistical Treatment of Animal Experiments [No. 1,

area, and should one particular treatment gain an ascendancy over the

others, it meant the introduction at a certain stage of an extra

animal. It was difficult, too, to fix in advance an integral number of

cows which could be depended on to graze the area effectively, while

mnixed grazing brought in other problems, and animals had a way of

falling sick and having to be replaced.

By that time satisfactory methods of crop experimentation had

been devised by Fisher, and the randomized blocks and Latin-square

lay-outs were becoming familiar, together with the essential principles

of good experimental design illustrated by these arrahgements. It

was natural, therefore, that animal experiments should be studied,

and criticised, in the light of these principles. From this point of

view the experiment I have just described was not satisfactory, since,

although live-weight gains were recorded for each cow on the plot,

there was no real replication of plots. The variation in live-weight

gain between cows on the same plot could not safely be taken as a

measure of the variation between cows in different plots; further,

the plots were bound to differ in inherent fertility, and a superiority

in pasture growth, as measured by its effect on animal growth, could

not be ascribed solely to the greater effectiveness of the particular

slag in supplying phosphatic fertiliser.

When it was suggested that the randomized block technique

should be applied to animals, various difficulties were raised. Thus,

it was stated that animals were more variable than field plots, too

variable, in fact, for small differences in growth to be detected.

Further, if this variability was to be reduced, all animals in one

experiment ought to be offspring of the same parents, and ought to

be of the same age and weight at the start of the experiment. This

limits the number of animals in a single experiment to the number of

young born at any one time, and even here it may be impossible to get

many of comparable weight. Thus the experiment can only be a

small one, and even if the standard error of a single animal's live-

weight gain is reduced by such a process of selection, the standard

error of the mean of any treatment will not, in general, be small,

beeause of the limited number of animals available. The expense of

animal experimentation is bound to limit the numbers capable of

being dealt with, thus keeping down the number of treatments to be

tested, and making an unsatisfactory experiment generally, since

while there may be several thousand unit plants of wheat in a single

plot of a cereal experiment, and several hundred unit roots on a plot

of potatoes or sugar beet, the unit with animal experimentation will

often be the single animal.

For these reasons progress in the direction of laying out animal

experiments which would permit of adequate statistical examination

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] WISHART-Statistical Treatment of Animal Expertiments 3

of the results has beell slow, and experimelnts have usually, been of a

very simple liature, comparable in fact to the duplicate plot experi-

ments that were in vogue prior to the randomized block era, except

that there was not always duplication. There have been exceptions

in the case of small animals; some of you may remember hearing

from Mr. Hartley the details of a fairly elaborate experiment on

poultry, at a recent meeting of this Section. I turn now to the series

of nutrition experiments on pigs- begun at Cambridge some two or

three years ago. These will serve as well as anything to illustrate

the methods involved, and they are experiments that I happen to

know something about. There may be others present who can deal

with cows, or sheep, or by contrast with small-scale animals such as

rabbits, or even rats and mice, while the point will not be lost sight of

that experiments on human nutrition deal with that very variable

quantity, the human animal. At an early meeting of this Section we

were given a contrast betweeni an experiment on rubber in Malaya

covering 2,ooo acres and one at East Malling, for which the total area

was only i/iOO acre. There may well be conitrasts as great on the

animal side.

with. Relatively pure stocks are obtainable, and large famiilies are

the rule. Thus, particularly at a time when the Boards for Pig

Marketing and Bacon Development were set up, it was natural that

experiments should be unidertaken, not only to investigate the most

economical method of producing a satisfactory bacon pig in this

country, but also to study the experimiiental methods themselves.

The usual method has been the group-feeding trial, and I shall

illustrate this from an experiment set up alongside the individual-

feeding trial which will be described later. Three rations were tested:

A, a diet containing the amount of protein usual in farming practice,

B, a diet with an excess of protein, and C, a diet with a protein excess

about twice, by weight, that of B. Three pens were used, one for

each treatment, and each contailled ten pigs, both sexes beilng repre-

sented in the same proportions. The produce of a number of litters

was distributed as evenly as possible over the three pens; for example,

if six pigs were available from one litter, two would be allocated to

each pen. Care was taken to see that the animals were as homo-

geneous as possible, and approximately of equal weight at weaning

time. There were, of course, certain variations, but the three pens

were evened up so that the average weaning weights were approxi-

mately the same. After the pigs had been given a little time in which

to settle down to their rations, they were weighed individually, then

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

4 WISHART-Statistical Treatment of Animal Experinments [No. 1,

standard tables, until i6 weeks had passed. By that time some of

the pigs had reached 200 lb., and had to go to the bacon factory.

The remainder were kept until they, in turn, had reached 200 lb.,

but the experiment proper ended when seventeen consecutive weekly

weighings had been recorded for each pig, all covering the same

calendar period. The effect of the rations was measured by the

average live-weight gain per pig over the i6-week period, the figures

being 142-5, 141V0 and 132-7 lb. for rations A, B and C respectively.

It is doubtful how far any kind of statistical analysis on the individual

live-weight gains is valid. If, however, we may assume that we can

regard the ten pigs in any one pen as replicates of one another, then

it appears that the drop in live-weight gain with increasing protein

percentage in the ration is not significant. The standard error of

each pig's gain is 17-3 lb., or 12-5 per cent. of the mean, and that

of the above three averages is 5.5 lb., or 4-0 per cent. A point of

importance to the experimenter is the comparison of the efficiencies

of meal conversion with the different rations. In the present case

only the total meal consumption for the ten pigs is known, since the

food was provided in bulk to each pen, and while the amount of meal

per lb. of live-weight gain can be worked out for A, B and C separately,

no figures can be got for individual pigs, and no statistical analysis,

however invalid, is possible.

The standard error obtained is as low as we are likely to get with a

group-feeding trial, for the experiment was conducted with all

possible care. We thus see that if we are to be able to detect differ-

ences of the order reached in this experiment, something more

precise is necessary. For some time individual-feeding trials had

been advocated, and the opportunity was now taken to make a

comparison of the two methods, particularly as the group trial was

there to convince the farmer, who might well distrust the results of

trials carried out in a way that was not normal farming practice, but

who would probably be convinced if both methods gave the same

results.

Individual-feeding trial

From the litters of each of five sows were selected six pigs, three

being hogs and three gilts. The six from any one litter were run

together in a pen, one hog and one gilt being given ration A, another

pair ration B, and the third pair ration C. Mechanical methods were

adopted to segregate the pigs at feeding time into small individual

pens, so that each pig had undisturbed access to its day's ration.

The experiment consisted of five pens, or thirty pigs in all. Weighings

were recorded as in the group-feeding trial. The experiment is of

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] WISHART-Statistical Treatment of Animal Experimtents 5

the randomized block type (for the rations were randomized in the

feeding-boxes), each block consisting of the pigs from one litter.

Thus pen (or litter) differences could be eliminated by the method

of the analysis of variance, showing the possibilities of this type of

experiment in cases where more pigs are needed than are available

fromn a single litter. The analysis follows the usual lines, and it is

possible to examine the effect of food, a possible sex difference, and

the interaction of these two effects, on any one of the variables.

These were: live-weight gain, meal consumption (known for each

pig), and a great number of post-slaughter measurements on the

carcase whose examination would reveal, if any, differences in

the conformation of the bacon pigs produced in consequence of the

differences in the composition of the food.

In the case of live-weight gain the differences were small, the

figures being I5-0, I46-4 and I42-6 lb. for A, B and C respectively,

with a standard error of 3-0 lb., or 2-I per cent. One very satisfactory

feature of the experiment was the fact that the accuracy was much

greater than that of the group trial; the standard error of each pig's

live-weight gain was 9-6 lb., or 6-5 per cent. of the mean, compared

with I7 3 lb., or I2-5 per cent. in the other case. The same story

is told by the means in both cases. Small as is the standard error,

the differences between A, B and C are not, however, significant on

the analysis of variance test, even when we note that nearly all the

sum of squares for the 2 degrees of freedom for food is contained

in the single degree of freedom component of " principal effect ",

which measures the drop from A to C. There is a way, however,

in which accuracy can be gained. The growth of a pig in any one

week obviously depends to a considerable extent on its weight at

the beginning of the week. Thus differences in initial age and weight

at the beginning may, and usually will, affect the comparisons. We

have no difficulty with age in the present case, but the limited

amount of experimental material makes it impossible to have the

six pigs in any one pen of the same initial weight. But this does

not really matter, as by means of the analysis of covariance we can

examine food and sex differences in live-weight gain after correction

for initial weight-i.e., adjusted to what they would have been,

knowing the regression of live-weight gain on initial weight, had all

pigs started at the same weight. When this was done it was found

that the food differences were still insignificant on the 2 degrees of

freedom, but when the single degree of freedom " principal effect "

was isolated, this was significant at the 5 per cent. probability level.

Thus, by a process of " squeezing," the drop in live-weight gain

from A to C was declared to be significant; the drop was 5-6 per cent.,

which figure had a standard error of 2-4, compared with 2-9 before

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

6 WISHART-Statistical Treatment of Animal Experiments [No. 1,

adjustment for initial weight (without adjustment the drop was 5-7

per cent.). In neither case was there a significant difference due to

sex. It is of interest to observe that live-weight gain was signi-

ficantly correlated with initial weight (r = + 0.60), while in the

case of the group trial the correlation was not significant (r = + 0.36),

due perhaps to competition at the feeding-trough, the weakest

going, possibly literally, to the wall. This is a further point in

favour of the individual trial.

These results, together with the results of examination of the

rest of the data, were published two years ago.' The experiment

was found to be quite workable without too much trouble, once the

piggery had been built for the purpose. It was the pioneer of a

number of trials which have been conducted since, limited, of course,

to the same total number of pigs and the same number of treatments,

although the treatments themselves have varied. As far as I was

concerned, the figures then left the nutrition experimental stage and

entered on a new stage-that of the statistical study of growth.

As it is the statistical treatment of such experiments that is the

subject of our discussion (though this involves, naturally, the

question of lay-out), I shall ask you to bear with me a little longer

while I explain what I have since done.

I have always thought it a pity that more use was not made of

the regular weighings of the animals in such an experiment as this.

Ten years ago I fitted straight lines by the method of least squares

to the weights of the cows in the basic-slag trial, in the belief that

this would furnish a more accurate measure of the growth than the

difference between initial and final weights, and I turned now to the

study of the growth curves of the pigs. These appeared on graphical

inspection to be parabolic in character, with an upward curvature

-i.e., the weekly gains rose up to practically the end of the experi-

ment. Now, had the gain in weight been proportional to the weight

at the beginning of any one week, the logarithms of the weekly figures

would have been linear, but plotting revealed that this was not so,

for the graphs showed a downward curvature. I therefore began

with the fitting of second-degree parabole to the actual weights,

figures more easily interpretable by the practical man than the

logarithms, having ascertained graphically that such a parabola

appeared to be a good fit to the data. There were now two variates

for each pig, g, the average growth rate in lb. per week, and h, pro-

portional to the rate of change of growth in lb. per week per week.

Analysis of variance showed that on g the food difference was signi-

ficant on the single degree of freedom for " principal effect ", repre-

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] WISHART-Statistical Treatment of Animal Experiments 7

The sex difference was not significant. Analysis of covariance on

initial weight w showed a further-gain in accuracy, for food differences

were now significant on the 2 degrees of freedom, and the sex difference

was found to be significant, the gilts having the greater mean growth

rate. The correlation between g and w was + 0-65. Although the

trend for A, B and C in the case of h was in the same direction as with

g, food differences were not significant, but there was a significant

sex difference, in that gilts had a greater rate of change of growth

rate than hogs. Covariance of h on initial weight showed no change

in the effects, the correlation being very small (r = + 0.04).

These results were given in a paper published in Biometrika this

year,2 but I have since returned to the attack, more especially as

further examination revealed that the cubic term of the curve fitted

to the weekly weighings, which I shall call i, was significant, while

the quartic term was not. Graphical examination of g and h against

w, and of h against g, revealed nothing abnormal. It seemed worth

while to see how far h appeared to depend on g, but analysis of

covariance showed that the two constants were not significantly

related (r + 0.20). In spite of this, however, the analysis of

residual variance after correction for g showed that the sex effect in

h fell below the 5 per cent. significance level. It would appear,

therefore, that it was the higher g in the case of gilts that was the

cause of the higher h, although the sex effect was not significant on

g alone. But there was an effect of initial weight on g, and this

suggested that the partial variablesg . w and h . w should be examined.

The following are the figures for the totals of IS pigs, for hogs and

gilts separately:

w q h

Hogs ... ... 617 138 2.43

Gilts .. ... 586 141 -9-73

rgh. V ( gh 2) approxim

0-20 0 26.

V\1 - 0.65 2)-

Thus it would appear that the significance of the sex effect for h

corrected for both g and w should be less than for h corrected for g

only. This was borne out by a multiple covariance analysis, which

reduced the sex effect to complete insignificance, although the

multiple correlation was not significant (R = 0 23).

Analysis of variance of the cubic term i showed a non-significant

sex difference, but very marked differences due to food treatments,

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

8 WISHART-Statistical Treatment of Animal Experiments [No. 1,

far more significant than anything hitherto reached. The figures for

the averages of ten pigs each are:

Standard.

A B c Mean Error

Mean i ... -000959 -000610 -000248 -0O00606 0 00109

Per cent ... 158.3 100-7 40.9 1000 18.0

between B and C, are significant. Analysis of covariance of i on w

added nothing to the story, although the correlation was just

significant (r -0.43).

The result of all this is that we are now in a fair way to obtaining

a complete picture of the growth curves of the pigs under food and

sex differences, and can say with confidence that they are significantly

different. For the five hogs and five gilts, one from each pen,

given over to each of the three food treatments, it was possible by

working out the averages of the constants of the fitted curves-

namely, the mean and the linear, parabolic and cubic terms-to con-

struct curves for the " mean pig " in each case. These constants

are given in the following table, together with their standard errors,

obtained in each case from the analysis of variance tables. Note

that they are not the quantities g, h and i hitherto discussed (which

are the B, C and D of Fisher's treatment of orthogonal polynomial

fitting in Statistical Methodsfor Research Workers), but the quantities

a,l-12g, a2 = `ih I i, and, of course, ao, which are the imimiediate

result of the application of Aitken's method of polynomial fitting,3

and can be used directly, as indicated by him, for calculating the

curves.

Coefficients of Curves (actual weights)

aO a, a. al

Hogs A ... ... 109.47 4.803 0-0553 -0-00273

B ... ... 104.48 4 495 0.0568 -0-00146

a ... ... 106.97 4-478 0.0497 -0-00052

Gilts A ... ... 106X88 4-846 0.0641 -0-00302

B ... ... 108.71 4.793 0*0587 -0-00221

a ... ... 101.98 4.498 0 0593 -0-00096

Mean ... ... 106.42 4.652 0.0573 -0-00182

Standard Error ... 3-87 0.144 0 0037 0.00046

or 3-6% or 3.1% or 6.4% or 25.5%

The six growth curves are shown in the diagram, each being

displaced relative to the one on its left by the same constant distance

at the start, in order to avoid overlapping. The points on the

diagram are the weekly averages of the weights of the five pigs for

each sex and treatment. The curve was not directly fitted to these

points by least squares, but was drawn from the averages of the

constants fitted to the individual pigs.*

* It will be realized, as pointed out by Professor Fisher in the discussion,

that the curve so obtained is in fact identical with that obtained by a least-

squares fit to the weekly averages.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] WISHART-Statistical Treatment of Animal Experiments 9

difference that are brought out by statistical treatment of the growth

constants. Comparing the food treatments A, B and C, on the average

of hogs and gilts, it is seen that A has the smallest initial weekly

gain and C the greatest, B being intermediate. The last weekly gain

is, however, the same for all; A has curved up more than C, and has

reached its point of inflexion before the experiment ceased, while the

C curve behaves more like a parabola all through. In case it may

be thought that different initial and final weights account for these

differences, it should be added that interpolation of the figures for

180

160

140

120

80

60

40

A 8 C A B C

HOGS GILTS

FIG. 1.-Growth curves of the "mean pig" for each treatment and sex

(actual weights).

the week after reaching 42 lb. and for the week before reaching i8o lb.

shows the same relative effects. Comparing hogs with gilts, averaging

for food treatments, we find that hogs start off with the greater

initial weekly gain and end with the smaller final weekly gain.

Hogs were, however, heavier at the beginning, and correcting for

this as before by interpolation, we find that the initial gains become

equal, though the final gains show no change. The curves showing

the two extremes in growth are those for hogs under the C treatment

and for gilts under the A treatment. These are side by side in

the diagram.

Finally, because three constants have been found to be necessary

for the description of the growth of the pigs on actual weights, it is

interesting to see whether any economy in this respect is possible if

logarithms are used instead. If growth differences can be elucidated

by examination of one or two constants of the curves fitted to the

logarithms, that will represent a distinct advantage. Analysis of

the logarithms of a sample pig of the A treatment showed that

the linear and parabolic terms were significant, while the cubic

B2

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

10 WISHART-Statistical Treatment of Animal Experiments [No. 1,

and quartic terms were not. Accordingly the first two terms, which

we may call g' and h', were fitted to all pigs, and the usual analysis

of variance was worked out in both cases. The table for g' is as

follows, g' being calculated from the logarithms taken to two

places of decimals, and treated as whole numbers:

Analysis of Variance-g'

Variation D.F. Sum of Squares Mean Square

Sex ... ... ... 1 0-2811 0*2811 1-1435

Food ... ... ... 2 0-2101 0-1050 0.6511

Interaction ... ... 2 0-0329 0.0164

Pens ... ... ... 4 0.9115 0.2279 1.0386

Error ... ... ... 20 0.5710 0*02855

The sex difference was very significant, the mean for gilts being

distinctly higher than that for hogs. Further, the food differences

were significant, the trend being in the same direction as in the case

of the g values. The following tables summarize the results, apart

from the significant pen differences:

Summary of Results-g'

Hogs Gilts M[ean Standard Error

Mean g' . ... 4-091 4-285 4.188 0.044

Per cent. 97.7 102-3 100.0 1.06

Mean g' ... 4301 4.163 4.101 4*188 0-053

Per cent. ... 102.7 99.4 97.9 100.0 1.28

clear-cut nature of the results, it was evidently not necessary to take

account of initial weights in a covariance analysis,* and the final

examination undertaken was the covariance of h' on g'. Contrary

to our experience of h and g, the regression was found to be significant

(r = -0 62), but adjustment of h' for g' did not alter the conclusions

already reached.

We conclude, then, that the simplest statement of the results of

statistical examination of this body of data is that treatment A has

given the largest mean proportionate growth rate, and C the smallest,

B being intermediate. Gilts have shown a greater mean proportionate

growth rate than hogs. All show a falling off to about the same

extent in proportionate growth rate as the period advanced. The

* The correlation coefficients for g' and h' with the logarithms of initial

weights were - 0-81 and + 0-47 respectively (both significant). The only

change on correction was that pen differences were now insignificant, showing

that differences in average growth constants for the pens were due to different

initial weights.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] WISHART-Statisticat Treatment of Animal Experiments 11

for the six " mean pigs ", in the same form as the previous table:

a.' a. a, a3

Hogs A ... ... 199 49 2.122 -00424 -0-000021

B ... ... 197*47 2-034 -0-0368 -0*000712

C ... ... 198-72 1 981 -0-0368 -0-001241

Gilts A ... ... 197'70 2-179 -0 0397 -0-000121

B ... ... 199.23 2-129 -0 0425 -0-000018

C .. ... 195*56 2-119 -0 0389 -0-000970

Mean ... ... 198.03 2 094 -0Q0395 -0 000507

Standard Error ... 1.66 0-038 0-0025 -0-000357

or 0.8% or 1-8% or 6-2% or 70-4%

The six growth curves calculated from these constants are shown

in the secoind diagram, together with points which represent the

12 30

c^o

~~~~~~~/ /

A B C A B C

HO8GS GILrS

(logarithms of weights).

weekly averages of the logarithms of the weights for each sex and

treatment. The graphs bring out the differences in average slope

as between A, B and C on the one hand, and between hogs and gults

on the other.*

It is not suggested that all the above calculations are necessary in

any similar experiment. In fact, had we started with the logarithms

instead of the actual weights, it is unlikely that the latter would have

had the intensive examination which has been given to themt. But

the figures are presented as an illustration of attempts that may be

* Further examination has shown that the cubic terms fitted to the B and

a hogs, and the C gults, are significant. The values of a3' have been added to

the table since the meeting, and the diagram has been re-drawn. It would

appear, therefore, that working on the logarithms does not necessarily save

computational labour to the extent hoped for, if a comaplete specification of

growth has to be provided.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

12 WISHART-Statistical Treatment of Animal Experirments [No. 1,

tion in a more complete way than has usually been done, by means

of relatively few growth constants. The imposed differences of

treatment in this experiment are small, but I understand that nutri-

tionists are interested in the change of form of the growth curve in

more extreme cases, with a view to relating these either to the food

supplied or to the subsequent conformation of the carcase. It may

be that similar methods to those outlined will prove useful in such

problems. They are, at any rate, open for discussion.

I shall conclude by referring briefly to a development which may

be taken to be the direct outcome of the Cambridge experiment, and

for details of which I am indebted to Mr. Boaz. The ordinary centre

can only experiment with a limited number of pigs, and must resort

to group feeding. A number of co-ordinated experiments have,

therefore, been arranged to be carried out at research stations, farm

institutes and agricultural colleges in Great Britain and Northern

Ireland. Three treatments, one of which is usually a control, are

allotted to three pens, each having five or six pigs, which are group-

fed. Sets of three pigs are taken, from the same litter and of the

same sex and roughly the same weight when I2-I6 weeks old. Each

set is regarded as a " block ", and the pigs in the set are allotted at

random to the three pens under the different treatments. Thus each

pen contains one pig from each set. Regular individual weekly

weighings are made until the pigs reach bacon weight, and later

carcase measurements are taken at the factory. Statistical analysis

at each centre is of somewhat doubtful validity, but each centre can

be regarded as a replicate, and the data can be analysed as a whole.

At some centres there is effective replication of the sets of three

pens; at Harper Adams College, for example, there will be in the

present experiment eighteen pens of five pigs each, testing three

treatments in six fold replication. The results of this experiment

will be awaited with interest.

The CHAIRMAN said that it was not the Society's custom when a

discussion had been opened to move a formal vote of thanks to the

opener, so he would take the opportunity of expressing his thanks

References.

2 Wishart, J., 1938, Biometrika, 30, 16-28.

3 Aitken, A. C., 1933, Proc. Roy. Soc. Edinb., 53, 54-78.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] Discussion on Animal Experiments 13

address. He particularly wished to call attention to Dr. Wishart's

wise choice of subject. Animal experimentation was now awakening

from a condition of backwardness relative to agronomic experiments,

and Dr. Wishart had put before the meeting two or three carefully

described and simple experiments, both of the cruder and of the more

advanced types, upon which he would probably welcome detailed

enquiries.

Animal experimentation, which was going forward in many

directions, was a very wide subject. There were cases such as those

with which Dr. Wishart had dealt wherein the animal was regarded

primarily as a commercial product and experimentation was aimed

at producing that commercial product most economically. Dr.

Wishart had alluded in his paper to man as a subject of animal

experiment. The more earnest advocates of nutrition in these

times seemed. to believe that the juvenile human should be brought

up like a little pig, perhaps fearing that otherwise he would be brought

up like a little wolf, but the material difference, as between human and

animal experiments, was that the pig was expected to be turned

shortly into bacon, whereas it was hoped that the child would be

capable eventually of becoming self-supporting, and perhaps also

capable of self-control.

Finally there was that very large realm of animal experiment

in which the animal was used for purposes of assay, and the reaction

of the animal was a means of, for example, standardizing insulin or

carrying out operations of a comparable character.

He had been exceedingly glad to hear Dr. Wishart, from the

evidence he had given, reinforcing the practice of individual feeding,

which had been advocated now for many years by Crampton in

Canada and was, he believed, becoming universally adopted by

animal experimenters. He wished also to call attention to Dr.

Wishart's judicious treatment of growth curves, and particularly to

the fact that many statisticians, when faced with a growth curve,

would, at least until recent years, have been tempted to analyse the

variation in weight at a given age and later to try to combine the

evidence obtained at different ages, which latter process was hope-

lessly encumbered by the fact that, using the same animals, the

evidence to be combined was far from independent. In the method

which Dr. Wishart had used, each individual animal was made to

supply a series of constants which adequately described its growth

curve, and those constants might be treated like any other experi-

mental data, and rendered valid by the random sampling of the

material.

He felt that in spite of the lucidity of the exposition, something

must have been kept back when Dr. Wishart, in speaking of those

growth curves shown in the diagram, said, " Note t-hat the curve is

not fitted to these points by least squares, but is the one determined

from the averages of the constants fitted to the individual pigs."

He could not see how that could be. He would expect to reconstruct

the least-square solution for the average weights by averaging the

least-square curves for the individual pigs.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

14 Discussion [No. 1,

least-square solution had been re-constructed in the process, and

the wording ought to be altered to make this clear.

address by Dr. Wishart on a very important subject.

It had been said that it was all very well to use a good statistical

design with a field crop, but that hopeless difficulties were encountered

if the same thing were attempted on animals. He doubted whether

that statement were true. It was true that animal experiments

required more continuous attention. But against that was the

advantage of being able to repeat experiments at decidedly shorter

intervals, since with such experiments-and he had particularly in

mind pigs and the smaller animals-one need not wait for the next

season to come round.

Again, with feeding trials and similar experiments it was not

necessary to start the whole of an experiment at one time. Each

block could be started when a suitable batch of animals was forth-

coming.

Once a resort was made to individual feeding, a large number of

units could be treated separately, and that removed what had in

the past been one of the greatest difficulties in the way of adequate

statistical design and, in particular, of introducing anything in the

nature of a factorial experiment.

In general, therefore, he would say that animal experiments of

the ordinary type tended to be just as simple as agronomic experi-

ments, and definitely more so than the sort of problems which were

encountered when dealing with long-term experiments in agriculture

concerning crop rotations and so on. On the other hand, it was

certainly true, because of the larger size and cost of the individual

unit, that animal-breeding was in general more difficult than plant-

breeding.

Much had been said, at various times, about the high variability

of animals. Certainly a single animal-a pig, for instance-was

variable, but from Dr. Wishart's figures, and from similar figures

obtained in experiments at Rothamsted, it would be seen that in

fact there was not much to choose in this respect between one pig

and one agricultural plot. They both tended in a well-designed

experiment to give an experimental standard error of the order of

iO per cent.

The utility of factorial design in animal experiments was only

now beginning to be appreciated. As an example of this type of

design he might instance an experiment on four factors carried out

at Rothamsted on the feeding of young pigs. The factors were:

large versus small amounts of green food, rationing versus ad lib.

feeding, comparison of pigs permitted to exercise in a field each day

versus those confined to pens, and coarse versus finely ground meal.

This gave i6 treatment combinations. of which 8 were assigned to

each pen or group of pens, the four-factor interaction being con-

founded. 64 pigs in all were used, each block of i6 being started

at an interval of about four weeks, Individual feeding was used,

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] on Animal Experiments 15

information was provided on all four factors simultaneously, and at

the same time possible interactions between the various treatments

were investigated. Each pig was in effect used four times over,

with a corresponding gain in overall efficiency, quite apart from the

information on the interactions.

Such an arrangement was, of course, only possible if a reasonably

large number of experimental units were available. This was one

of the great merits of individual feeding. Another advantage was

that proper estimates of experimental error were available from

even a small number of pens. The comparison of the individual

weights in a group-feeding experiment could never furnish such an

estimate, not only because all the pigs in the same pen might suffer

some common damage, such as disease, but also because of competi-

tion within each pen. Lastly, individual feeding might be expected

to increase accuracy considerably, since it eliminated competition

at the feeding-trough.

There were other possibilities in the design of animal experiments

which were not available in agronomic experiments. One of these

was the rotation of treatments on the same animals. Mr. Wake

Simpson had recently given a paper to this Section on certain

experiments he had carried out on school-children, in which he had

made use of this principle.. The objection to rotation of treatments

was that the effects of the treatments might persist from one period to

the next, so that the results might be complicated by residual effects.

Use might also be made of concomitant information such as

initial weights, inilk yields over a preliminary period, and so on.

The statistical process here involved was known as the analysis

of covariance. There were, in fact, many possible ways of increasing

the accuracy of an experiment: it was the job of the statistician

to incorporate these in the design, and not to attempt to use them

when the experiment was completed, but whatever refinements of

design were adopted, proper replication and randomization were

essential. It might seem unnecessary to emphasize this, but a

most cursory examination of the literature would show how often

these cardinal principles were neglected, and how often through their

neglect the whole of the work was rendered valueless.

In conclusion, a brief description of a grazing experiment at

present in progress at Rothamsted might be of interest. The

problem was that of evaluating the residual effects (on the pasture)

of feeding cake to grazing cattle. It was a difficult problem experi-

mentally, and had been made more so by the agriculturists because

they insisted that practical grazing conditions had to be reproduced,

and that to do this each plot must be at least five acres in area.

In the Rothamsted experiment there were nine five-acre plots in

three blocks of three plots. In each year one block was grazed by

cattle, those on one of the plots being fattening bullocks to which

cake was fed. In the subsequent winter one of the other two plots

of the block received nitrogenous fertilizer. On the third plot

nothing was given. In the two subsequent years mixed grazing was

adopted, with the intention of measuring the effect of the treatments

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

16 Discussion [No. 1,

envisaged that samples of the grass might be cut and the yields

evaluated quantitatively. The remaining two blocks were treated

similarly at one-year intervals, so that all three phases of the rotation

were represented.

Whether the experiment would ever be capable of providing the

hoped-for results remained to be seen-Dr. Yates had grave doubts

on the matter. But it was at least interesting to note that the

variability of the animals in the first year (which was a bad one

from the grazing point of view) was by no means as large as was

feared. From the comparison of the increases of animals on the same

plot the standard error per plot due to the variability of the animals

was found to be only about 6 per cent. of the increase in weight.

This error was of course not cumulative from year to year. Another

interesting result that had already emerged was that mixed grazing

was more efficient from the point of view of live-weight increase

than grazing by cattle alone.

MISS TURNER said this was her first appearance in the Society,

and she expressed her thanks for being allowed to be present. She

wished to speak on some problems which might be likened to the

long-term experiments of which Dr. Yates had spoken. Her own work

had been done in Australia, and related mainly to sheep. In

Australia the natural pastures were deteriorating, and one method

of improving them was to plant English grasses and fertilize them

annually. Some graziers had raised the objection that; although

this increased wool production, it caused coarsening of the wool,

with consequent decrease in value. It was necessary, therefore, to

find out what effect the improvement of the pasture had on the wool,

and whether such an effect was cumulative. It was also important

to determine the effect of improved nutrition, such as that provided

by improved pastures, on lambs whose growth had been checked at

various stages by feeding on poor pastures. Both these might be

described as long-term experiments.

Another point was the use of the group trial. Miss Turner

agreed that individual feeding was essential for a laboratory trial,

but thought another kind of trial was necessary in Australian

conditions. Pigs could be fed in pens either individually or in

groups, and when thus feeding them in groups one was approaching

quite closely to actual husbandry conditions. But feeding sheep

in pens was getting nowhere near actual farming conditions in

Australia, and the grazier was far more likely to be convinced by a

trial which did approach such conditions. Experimenters had to

convince both themselves and the grazier by translating any trial

from laboratory conditions to the field. She would describe one

large experiment with cattle which was being carried out by the

Australian Commonwealth Council for Scientific and Industrial

Research.

A certain cattle disease in Queensland was known to be a deficiency

disease, and preliminary experiment had shown that the feeding

of certain supplements was successful in preventing it, 13ut it was

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] on Animal Experiments 17

into the field. About 3oo animals, varying in age, sex and breed,

were collected from various sources. It was decided to run a

randomized block experiment, arranging the animals so that in each

block they were uniform for age, sex, breed, the station from which

they were purchased, initial body-weight, growth curve during a

preliminaryperiod, and size. Body-weight always presented a problem,

since the weight of the same animal might vary at differenit times

of the day, and the animals were therefore run together for nine weeks,

being weighed weekly. The initial weight was taken as the mean

of the nine weighings, and the growth curve over the nine weeks was

also considered. Size was measured in terms of height, length and girth,

since if body-weight alone were considered a large cow in poor con-

dition would not be differentiated from a small cow in good condition.

The main trial consisted of three groups, two receiving supplements

and one being the control. It was hoped that in addition to analysing

any mean responses to treatment it would be possible to sort out any

sex differences or breed differences. In the preliminary experiment,

which had given an indication of successful treatment, no control of

variation such as the use of randomized blocks had been attempted,

and by their introduction the difference necessary for significance

had been considerably reduced.

dealing with a variety of animals for experimental purposes but he

himself was particularly concerned with dairy cattle of all ages. He

thought the question of whether to adopt group or individual

feeding for dairy cattle was still of controversial interest. With

mature dairy cows the matter was not important because it was

easy to treat the cow as an individual both in regard to intake of

food and output of milk, and live-weight, but it was sometimes

difficult to arrange individual-feeding experiments for young cattle,

and at Shinfield, until recent years, the group-feeding method was

always used. For experiments on the winter feeding of young

stock it was usual to let the animals run loose in a yard, half a dozen

or so together, and feed as best they could. During the last year

or so, however, individual feeding had been introduced, the

practical difficulties having been overcome by tying up the

animals during the day so that all their food was taken individually,

and letting them loose at night for exercise and so-forth. This plan

worked remarkably well, both from the practical and the experi-

mental point of view. The experiments were now of the random-

ized block type with as many treatments and replications as the

number of animals permitted.

Dr. Wishart had mentioned a comparison of the variability of

group and individual feeding. He could safely say that the standard

error of individual live-weights in young cattle on individual feeding

was less than half of what it was when they were on a group-feeding

system. With cattle, therefore, they showed perhaps slightly more

improvement in this respect even than Dr. Wishart had found with

pigs.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

18 Discussion [No. 1,

young stock was that it imitated practical conditions, but this

argument needed qualification. Under practical conditions a farmer

who had a group of animals in a shed, and found one of them being

badly bullied and half starved, would probably place the animal in

another pen. This would be good, or at least relatively good, farming

practice, but it would spoil an experiment.

He considered it paid to tie up animals for feeding during the

winter whether they were under experiment or not. An improvement

in these animals was noticeable over animals running loose, in mean

live-weight gain as well as in uniformity.

The amount of bullying of certain animals in group feeding

varied, and he strongly suspected that this bullying propensity was

influenced by nutrition. This would suggest that there were other

factors to be taken into account besides live-weight gain. The

grazing experiment which Dr. Wishart mentioned also brought to

mind all sorts of other factors. While he had tried to show that

individual feeding was a great advance from the point of variability,

he believed that some of these other factors could only be tested by

group feeding. It would be well, therefore, not to assume that

because individual feeding gave less variability than group feeding

the latter must be given up. Group feeding was still useful in

measuring certain factors.

He was interested in Dr. Wishart's growth curves, which appeared

to place the animal experimenter ahead of the crop experimenter,

because the former would now utilize the interim weights. Here he

would like Dr. Wishart's opinion. In a winter feeding-experiment

the cattle were weighed on three successive days both at the beginning

and again at the end of the experiment. During the course of the

experiment it was proposed to weigh the animals each month on

three successive days. One alternative might be to weigh them

one day each week. Which was the better method for use

with Dr. Wishart's group-curve technique? It was probably a

question of whether the greater accuracy of three weighings was

more important than the danger of missing the exact point of

inflection of the growth curve which might occur with monthly

intervals.

He had talked over the matter with a colleague of his who had

experimented with rats and it appeared that in rat-feeding experi-

ments the question of group feeding seldom arose. Individual

feeding was invariably used and was superior in every way for

nutrition experiments.

arose fairly directly out of the paper. The first was in connection

with design, and bore on the last section of the paper dealing with

Mr. Boaz's experiments. It was sometimes difficult to obtain

enough animals at any one centre, and it was therefore hoped to

conduct instead co-operative experiments with a rephcation at each

centre. That point was brought out in a discussion opened by the

late Mr. Gosset in 1936, who said that this might be all right if the

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] on Animal Experiments 19

variability in farming practice between centres might cause differ-

ences in the nature of interactions in the same way as in series of

field experiments. But in spite of that difficulty, Dr. Bartlett

thought there was still a greater chance of usino each cenitre as a

separate replication than in the case of field experiments. With the

latter all the nitrogen required, for example, might be obtainable from

the soil, whereas in animal-nutrition experiments differences were

likely to be of a smaller order, for the animals would always need

a definite ration with so muLch protein, so much carbohydrate, and

so on.

A difficulty that sometimes confronted the statistician in the

analysis of experiments (not only in animal experiments) was that

arising from several sets of observations. If the experimenter gave

his own instructions on what should be tested, there was hardly any

difficulty (though it might perhaps have to be pointed out that all the

information available in the experiment was not being extracted);

otherwise it was a separate problem in itself to consider what to do

with the data. Sometimes one could arrive at a compromise. In

long-term dairy-cow experiments at Jealotts Hill, the mean weekly

milk yield during the experimental lactation period was taken.

The farmer was mainly interested in this, and it also probably

gave a fairly efficient analysis of the differences between the lactation

curves due to treatment. In other cases one might have to depart

from the sort of thing the experimenter himself would select, and

try to reduce the observations for any treatment to one or two

constants. It was interesting to observe that by doing this for the

live-weight figures Dr. Wishart had established a clearer-cut difference

between the food treatments than had been obtained originally from

the straightforward live-weight increases.

There was still a slight statistical difficulty, even if one had cut

down one's observations to one or two constants, in that, strictly

speaking, it was not fair to isolate one analysis from others being

done at the same time. It might therefore be relevant to summarize

Dr. Wishart's analysis in regard to the joint significance of all the

fitted constants, which he would do by writing down the following

approximate but simple x2 table for the food effect (adjusted for

initial weights).

9 ... ... 2 7'12 13 28

h.g ... ... 2 0.45 0.51

i.gh ... ... 2 8 77 1.23

Note on the above Table.-Any item could have been tested exactly by the

z test. For testing the last line, see, for example, Proc. Camb. Phil. Soc., 30

(1934), 327-340 (339). For the approximate x2 test for the last line, as used

here, see Proc. Camb. Phil. Soc., 34 (1938), 33-40 (380). The other items were

chosen to make the table additive. It should be noted that (i) the x2 test is

leas accurate for these items, since the multiplying factor used to get the best

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

20 Discussion [No. 1,

x2 approximation does no

(ii) caution should be used in interpreting the significance of any partial effect

such as h.g (h adjusted for g), since the adjustment implies using covariance

with g, which is itself varying significantly between treatments. Nevertheless

the entire table still serves as a useful summary of the significance of the food

effect. It was amended to include the cubic term for the logarithmic analysis,

which was not completed by Dr. Wishart until after the meeting. The

significance level of the total x2 nearly reaches P 0-01 for the direct analysis

but is only about P = 0 02 for the logarithms. It is interesting, however, to

note the large contribution to x2 for the logarithms from the linear term. The

cubic term in this analysis was found just significant by Dr. Wishart, but does

not provide any extra significance when tested jointly with the other constants.

experiments, but from time to time he had examined figures obtained

at Harper Adams Agricultural College in connection with pig

experiments. It was of some interest to consider the results of what

might be called a uniformity trial, carried out some years ago by

Dr. Crowther. Fifty pigs were divided into five lots of iO pigs each,

in such a manner that the lots were made as closely comparable

as possible, attention being paid to breeding, weight, age, and sex.

The IO pigs of each lot were housed in two sties, the pigs in one sty

being made comparable with those in the other. Throughout the

experiments all the pigs were on the same treatment. Taking as

the criterion the total gains in live-weight, the effect of making the

sties comparable at the outset was, as might have been expected, to

yield a very high variance within the sties and a significantly lower

variance between sties. The variance between the sties in the lots

was not significantly different from the variance between lots. A

further point was that if the pigs had been distributed at random,

the difference between the means of two lots of iO pigs each could

not have been considered significant unless it had been greater than

I 2 per cent. of the meani.

It was considered that the most satisfactory method of conducting

a group-feeding trial in which all the pigs in a sty fed at the same

trough, was to regard a group of pigs as the experimental unit, rather

than the individual pig. The prospect of being able to test only

differences greater than I2 per cent. was not at all encouraging, but

the low variance between the means of lots of iO pigs in the uniformity

trial, giving a standard error of only 3 I per cent. of the mean,

suggested that the method adopted to make the lots as even as

possible at the outset might be used.

The further requirements after the groups had been formed were

(1) adequate replication to improve the accuracy of the treatment

comparisons, and, what was more important, to provide an estimate

of error by which to judge the significance of treatment differences,

and (2) to assign the treatments to the different groups at random.

For what he described as reasons of good husbandry, the practical

man preferred to keep the large pigs separate from the small pigs,

and attempts had been made to introduce replication through

groups of pigs of different sizes on each treatment. In one such

experiment 6o pigs were first divided into four groups according to

initial weight, while in all other respects the groups were made as

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

1939] on 4nimal Experiments 21

then divided into three sub-groups, each to receive a different type

of management. The pigs of the other size-groups were treated in

a similar manner. The standard error of a treatment mean in this

case was 4-5 per cent., based on 5 degrees of freedom, after eliminating

variation between size-groups and correcting for differences in initial

group weights by linear regression.

Soine caution was necessary in taking as an estimate of error the

interaction between treatment and size. In a similar experiment

where there were three size-groups and a comparison was made

between two levels of protein in the ration, it was found that the

superiority of the one ration over the other was smallest for the heavi-

est pigs and greatest for the smallest pigs, while it was intermediate

for the pigs of medium size. It was impossible to say whether this

was significant or not, because of the lack of replication of the size-

treatment groups, but it seemed to be generally accepted that in

proportion to its size the small pig required more protein than the

large pig.

A serious drawback to the success of group-feeding trials at a

single centre was the limitation of accommodation for experimental

purposes. In the majority of the experiments at Harper Adams

College the number of pigs had been between 30 and 40, but in some

cases as many as 6o pigs had been used. The consequences of this

limitation were that the experiments had been of a simple type, the

number of treatments and replications was restricted, and the estima-

tion of error was usually based on so few degrees of freedom that only

large treatment differences could be shown to be significant.

There seemed little doubt that group-feeding trials could serve a

useful purpose in pig-feeding investigations, provided the experiments

were properly planned and carried out on a sufficiently large scale.

One means of meeting the latter requirement and at the same time

extending the range of conditions under which treatments were to

be tested was to co-operate with other centres of a similar character.

A certain amount of co-operation of this kind had already taken

place, and Dr. Wishart had given some details of a co-operative

scheme which was at present in progress. Incidentally, the part

which Harper Adams College was taking in this investigation was the

largest pig experiment carried out at that centre. From what he had

learned regarding the arrangements the experiment at that station

would effectively provide its own estimate of error.

He had been particularly interested in Dr. Wishart's remarks

concerning correlation between initial weights and total gains in

weight. His observations from the limited data he had examined

were that when the initial weights ranged from 35 lb. to 6o or 70 lb.

there was a positive correlation which was well marked. When the

initial weights ranged from 6o lb. to go lb. there appeared to be no

correlation. Further, among a group of pigs where the initial

weights ranged from 42 lb. to 96 lb. the regression of total gain on

initial weight was not linear, the medium-sized pigs gaining slightly

more than the larger and considerably more than the smaller

pigs.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

22 Discussion on Animal Experiments [No. 1,

mend weighing on one day in each week instead of taking three suc-

successive daily weights in each month. The latter process would

give a more accurate determination at any given time, but if growth

curves were to be fitted the accuracy of each individual weight was

not so important as the number which could be obtained.

In conclusion, he expressed his gratification at the way in which

his paper had been received. He was glad that the discussion had

widened out and dealt with different animals, different methods,

and different problems; also that there had been a pleasing absence

of criticism of his own contribution.

This content downloaded from 14.139.194.12 on Fri, 30 Nov 2018 06:34:30 UTC

All use subject to https://about.jstor.org/terms

- MODELING AND OPTIMIZATION OF A MOBILE CROP RESIDUE DISINTEGRATORUploaded byIAEME Publication
- STATISTICS IN RESEARCHUploaded byelenammanaig493
- Basic Formulae for Mixed Effects Models in RUploaded byLuke Mceachron
- ac.els-cdn.com_S092401360200273X_1-s2.0-S092401360200273X-mainUploaded byBalu Bhs
- FELIX RESULT.DOCUploaded bydanieltettley
- Reducing Variability With Experimental DesignUploaded byTino LC
- BF1Uploaded byAdithya Prabu
- Research Plan TahongUploaded byFatima Kate Tanay
- l13_ancovaUploaded bydamysaputra
- Statistical Analysis and Optimization of Ammonia Removal From Landfill Leachate by Sequential Microwave_aeration Process Using Factorial Design and Response Surface MethodologyUploaded byHector Andres Cabezas
- 12Uploaded bymichsantos
- 21 Inter CroppingUploaded byHéctor Yendis
- Alternatives to Difference Scores_Polynomial Regression and Response Surface Methodology_.pptxUploaded byjohnalis22
- Tute Exercise 11 Solutions1Uploaded byTushar Mehta
- 15Uploaded byNetP
- LAB Tortoise HareUploaded byfnoschese
- Analysis of TRAI DecayUploaded byAaronmiguel27
- Example of Lab Report Mark SchemeUploaded byAtha Aillah
- 79-86Uploaded byyooo20
- 3169004_1073605809_Labreportguidelines-2018-2019Uploaded byESAU
- Optimization of Surface Roughness in CNC Turning of Aluminium 6061 Using Taguchi TechniquesUploaded byIJMER
- ESTIMATION OF ANNUAL RUNOFF IN INDRAVATI SUB BASIN OF GODAVARI RIVER USING STATISTICAL APPROACHUploaded byIJIRAE- International Journal of Innovative Research in Advanced Engineering
- lesson plan 2 epUploaded byapi-353310138
- resumesirsUploaded byapi-284599970
- MASTER SCORE FORM-PeKA.xlsUploaded byMildred John
- ANOVA PresentationUploaded byPratik Kulkarni
- L6_RCBDUploaded byTeflon Slim
- Gomez. Statistical procedures agriculltural.pdfUploaded byPepe Garcia Estebez
- Mercury Concentrations in Crayfish (Orconectes virilis) Tissues from Soft-water Lakes on the Canadian ShieldUploaded byronaldjameshall
- Written-report.pdfUploaded byAbdul Basith

- Fem Modal UpdatingUploaded byAnjneya Srivastava
- Introduction to Finite Elements in Engineering, 3rd Ed, t.r.chandrupatla_2Uploaded byRohitash Nitharwal
- Yuan_et_al-1998-Earthquake_Engineering_%26_Structural_Dynamics.pdfUploaded byAnjneya Srivastava
- 1-s2.0-S0888327016304964-mainUploaded byAnjneya Srivastava
- 200 Best Ques Nov With AnswersUploaded byAnjneya Srivastava
- Shokrani_et_al-2018-Structural_Control_and_Health_Monitoring.pdfUploaded byAnjneya Srivastava
- Shokrani_et_al-2018-Structural_Control_and_Health_Monitoring.pdfUploaded byAnjneya Srivastava
- new paperUploaded byAnjneya Srivastava
- Damage detection using response surface methodology in shear building.pdfUploaded byAnjneya Srivastava
- Damage detection using response surface methodology in shear building.pdfUploaded byAnjneya Srivastava

- export of canadian cattle oilers for nepalese finalUploaded byapi-340500050
- 07_Nutrition_Health Dairy.pdfUploaded byladir
- Dairy ProjectUploaded byVivek Tiwari
- Breeds of Livestock - Ongole CattleUploaded byrajsurangi
- Vaginas Depicted in Sumerian: a DatabaseUploaded byJennifer Ball
- TNV-604Uploaded byrohishaak
- Dairy Goat Farming PracticeUploaded byBang
- harvest-moon-walkthrough.pdfUploaded byRisa Ardiani
- Naresh Kadyan We Support Him to Contest ElectionUploaded byNaresh Kadyan
- Soya Bean HullUploaded bySo Nic
- SacredCow.pdfUploaded byGigiPetrea
- indiankanoon.org-1 Cwp No8548 Of 2012 vs Union Of India And Others on 4 July 2012.pdfUploaded byVaishnavi Jayakumar
- Inventory_management Aavin MilkUploaded bysrinivas2help8836
- Quality Control Manual for Cattle Feed PlantsUploaded byngothientai
- Amul DaiiryUploaded byMiteshPatel
- Sustainable livehihoods in Hills of Jammu and Kashmir: Cheese KalariUploaded byJatinder Handoo
- CLA Cattle Market Report September 19, 2018Uploaded byClovis Livestock Auction
- History of Shorthorn cattleUploaded byRobert Scarth
- tudor farmingUploaded byapi-288380015
- How to Start a Dairy Farm_ 15 Steps (With Pictures)Uploaded byRaja Sekhar
- COK Vegetarian Starter Guide: The Whys and Hows of Vegetarian EatingUploaded byVanessa
- Comparison of rumination activity measured using rumination collars against direct visual observations and analysis of video recordings of dairy cows in commercial farm environmentsUploaded byIvan Filipović
- Beef & Dairy 2016Uploaded byThe Standard Newspaper
- R-48Uploaded byNiteshMaheshwari
- The Use Of SexSorted Sperm For Reproductive.pdfUploaded byAndra Sabina Valeanu
- Beef Cuts and UsageUploaded byRahul Desai
- List of All Merick Tbls&ChrtsUploaded byMena
- Modern Animal HusbandryUploaded byJelena Jokic- Grobler
- WHA Halloween Harvest Catalog.web.2018Uploaded byholsteinplaza
- 220-231Uploaded byMalik Makmur