4 INTRODUCTION 



such as the tables furnish, serves to reveal the relations in which 

 the control animals from the respective laboratories stand to one 

 another, and thus permits a more trustworthy comparison of 

 the experimental results. 1 



Moreover in the course of routine work on the same colony 

 one cannot be sure that the animals retain during successive 

 years the same relations to the reference table values. For this 

 reason we have been following the custom of referring all meas- 

 urements to the reference tables and using the difference in devia- 

 tion shown by the controls and by the test animals respectively 

 as the measure of the modification experimentally produced, 



By using such a procedure in place of the assumption that 

 the control animals from the same colony remain similar the 

 experimental results obtained from year to year are made fairly 

 comparable with one another. 2 



But there is still another use of the tables which is perhaps the 

 most important of any. In all experiments on the relative 

 weights of parts or organs in which the size of the test animals 

 differs from that of the controls, we readily obtain by weighing 

 or measuring the differences for the entire animal. If however 

 we wish to determine whether the relative size (weight or length) 

 of the parts or organs of the test animals has been affected, we 

 find that this cannot be done by comparing the test and control 

 groups directly for the relative values of parts and organs differ 

 with the absolute size of the animal but it can be done by 

 reference to the tables in which the desired values are given ac- 



1 If a strain appears in which the length of the tail is on the average 4 per 

 cent below the reference table value then if we compared directly with them the 

 test animals which came from a strain normally in agreement with the reference 

 tables but which through experiment had had their tail length reduced by 3 

 per cent it follows that the test animals, though modified by experiment, would 

 still have relatively longer tails than the first strain. 



Consequently to compare with each other the results obtained from the two 

 strains, the deviations of both the controls and the test animals from the refer- 

 ence table values must be determined in both series and the differences within the 

 series be used for the cross comparison. 



2 The same principle and procedure as described in Note 1 applies to the treat- 

 ment of different series taken, for example from our own colony, at different 

 times. 



