TRANSACTIONS OF SECTION M. 745 
An Experimental Attempt to value a Feeding-stuff. 
In the 1909-10 and the 1910-11 cattle-feeding experiments an attempt was 
made to find the value of soya-bean cake in comparison with linseed-cake. Two 
series of lots of cattle numbering respectively 36 animals, treated alike in other 
respects, received 4 Jb. per head per day of each food for periods of about four 
months. Nearly seven tons of each feeding-stuff were consumed. The increases 
were as follows: 
Cwt. qrs. lb. Cost per ewt. 
Linseed-cake . . 84 1 24 37s. 84d. 
Soya-bean cake... 78 0 5 35s. 54d. 
The difference amounts to 2s. 3d. per cwt. L.W.I. in favour of soya-bean 
cake, or 25s. 3d. per ton of that food consumed. 
The linseed-cake cost 97. 5s., and the soya-bean cake 6/. 15s. When the 
values of their manurial residues, 41s. and 52s., are deducted, the net food costs 
144s, and 83s. respectively. When 25s. per ton, the increased productive value 
of soya-bean cake obtained in the experiments, is added the relative food value 
of soya-bean cake becomes 108s. Thus the food value of soya-bean cake was 
three-fourths that of the linseed-cake, and the purchase value, taking linseed- 
cake at 97. 5s., would be 2 (185s. — 41s.) + 52s. = 8/. per ton. 
2. The Feeding of Dairy Cows in the West of Scotland. 
By Professor R. A: Berry, F.1.C. 
Attention was drawn to the great diversity in the feeding of dairy cows as 
practised in the principal dairying districts in the West of Scotland. The 
feeding in (a) districts where cheese-making is mainly carried on, and (b) dis- 
tricts where the sale of milk and butter are the main objects, was considered from 
the point of view of economy in the supply and utilisation of the nutrients of the 
food. 
Much of the information contained in the paper has been obtained through 
the kindness of the supervisor and secretary respectively of the Milk Records 
Society for Scotland. The paper was only intended as a preliminary contribution 
to this important subject. 
3. The Probable Error of Agricultural Experiments. 
By Professor R. A. Berry, F.I.C. 
Attention was directed to differences in the probable error calculated from 
pig-feeding experiments. 
In an experiment carried out at the experiment station of the West of 
Scotland Agricultural College in 1911, in which seventy-six large white pigs were 
used with an average initial live weight of 77°6 lb. with sex equally divided 
and all fed on the same ration for fourteen weeks, the probable error of one 
animal was 12°1 per cent. of the live-weight increase. Calculating from the 
results of previous experiments extending over the years 1905-08, and choosing 
only those lots which were fed on the same or similar diets, numbering 102 pigs, 
with an average initial live weight of 97 lb., the probable error of one pig works 
out to 13°7 per cent. of the live-weight increase. Both sets of figures give practi- 
cally normal frequency curves. The differences mean that twenty-one or twenty- 
seven pigs are necessary to determine with any degree of certainty a’ difference 
of 10 per cent. between different foods, and that thirty or thirty-eight pigs are 
required to determine a difference of 10 per cent. in either direction. These 
differences, though not large, point, when calculating the probable error, to the 
advisability of taking into account age and weight of animal at commencement 
of the experiment, also whether the data are from one complete experiment or 
from several experiments extending over a number of years. 
Fifty female pigs in the latter experiment gave a probable error of 13°5 per 
cent. of the live-weight increase, and fifty male pigs 13°8 per cent. 
Wood gives about 14 per cent. of the live-weight increase on the probable 
error for cattle and sheep. His method of calculation is followed here. 
