X PROLOGUE 



So far as these data are based on observations from our own 

 laboratory they apply to a fairly uniform population of rats 

 somewhat modified from time to time by food conditions. Fur- 

 ther, it may be said that these rats show the characters com- 

 monly found in other laboratory colonies, but they represent 

 neither the best animals nor those in the ideal condition, nor 

 do they necessarily yield the values which will be found in 

 Albinos ten years hence. Yet despite these defects the data 

 and reference tables still have their uses. 



The data in the reference tables reveal the form of the growth 

 curves for the entire rat and its organs. In their essential 

 features these growth curves are the same for rats of various 

 sizes. The tables however have another and quite different 

 use. 



The case for the tables may be put in the following form. In 

 a given research the test rats are compared with the controls 

 and the differences from the controls give the amount of change. 

 Here the controls furnish base line values, so to speak, and by 

 them the values for the test rats are measured. In this in- 

 stance reference tables are a luxury. Suppose however that 

 two years later the same investigation is repeated and the changes 

 in the test animals are again determined in the same way. 

 When the two series of results are compared with each other it 

 is important to know whether the control rats in both series 

 agree in their characters. If, as is very possible, they do not 

 agree, then it is a question which of the controls should be taken 

 as the ultimate standard. This situation arises in all fields 

 where measurements are made, and it is met by having ac- 

 cepted standards to w^hich the individual measuring instruments 

 (here the controls) are referred. It may happen then that the 

 controls in the earlier research show a character ten per cent 

 above the table values, while in the later research, it is ten per cent 

 below. In any effort to collate the two series of results this 

 difference between the controls must be taken into considera- 

 tion. 



Such differences may not be alwaj^s significant as influencing 

 the experimental results, but on the other hand they may be of 



