EXPERIMENT STATION WORK 
retical truths will be the experimental results. Note, how- 
ever, that small lots may by chance be as near the truth as 
large lots. 
In practice two grave errors are made: First, conclusions 
are drawn from small lots compared with each other; second, 
conclusions are drawn from large lots compared with small 
lots. In the first case both may be off; in the latter case the 
small one may be off. Examples of the first error are to be 
found in the scores of contradicting breed and feed tests, that 
were published in the early days of poultry research. The 
second error is exemplified in the Ontario experiments in in- 
cubation, to which reference has already been made. 
Here is a further example of this error. From the fifth egg 
laying competition at the Hawkesberry Agricultural College 
in Australia, I copy the following: 
No. of Hens. Variety. Ave. Egg Yield. 
Grecia diarteyeiae ney Cuckoo Leghorn .............. 190.16 
OU cin eae etusieaee S. C. Brown Leghorn.......... 177.00 
BSB is tsctaca ves, chow S. C. White Leghorn...... oe L74,93 
MDs ds ciicesasteus eitess R. C. Brown Leghorn..........173.50 
| ee R. C. White Leghorn.......... 172.66 
182¢ uae oe ste Buff Leghorn .............6. 160.55 
Gieenid seastene Black Leghorn .............. 138.33 
The ranking of Cuckoo Leghorns as first is a chance hap- 
pening due to the small number; likewise the Black Leghorns 
had a streak of bad luck and received lowest place. To one 
not familiar with such work, the real significance of the table 
is that the S. C. W. Leghorns did the best work. A totaling 
of all other varieties gives 84 fowls with an average egg pro- 
duction of 170.5, which bears out the conclusion. As these 
birds were all kept in pens of six, we would expect to find 
the highest single pen to be White Leghorns, because, when 
compared with all other Leghorns, they have both the highest 
average and the greatest number. This accords with the fact 
that as the highest single pen is found to be White Leghorns 
with an egg yield of 239 eggs. 
The above illustrates another important phase of the laws 
of chance, which says that not only is the average likely to be 
nearer the theoretical average sought when the number is in- 
creased, but that the individual extremes will be more 
removed. 
184 
