from the average, even by a single bird, may in- 

 dicate ill health. A shift of probably as little 

 as a half million erythrocytes per cubic milli- 

 meter could be significant. Likewise, hemo- 

 globin values show a relatively narrow range 

 of variability, and a difference of as little as 1 

 gram per 100 cc. might be viewed with suspicion. 

 Hematocrit values do not as a rule vary more 

 than about 5 percent either way. The average 

 hematocrit values in table 17 are only 1 to 2 per- 

 cent higher than those given by Hamre and Mc- 

 Henry (1942a) except for the males. Buffy 

 coats were read in the Van Allen hematocrit tubes 

 with a low-magnification hand lens and as long 

 as there was no hemolysis the normal values usu- 

 ally did not vary more than ±0.2 percent. 



On the other hand, thrombocytes, total white 

 cells, and individual cell types often varied from 

 half to twice the average, and thus individual 

 readings do not mean much. This was found to 

 be the case whether the comparison was among 

 different individuals or among repeated bleed- 

 ings from the same bird. Palmer and Biely 

 (1935a) studied the variability of cell counts in 

 great statistical detail and concluded that when 

 careful attention has been given to technics the 

 fluctuations in normal erythrocytes can be re- 

 duced to 15 percent. In their data, the coefficient 

 of variability was generally quite low. 



This brings us to the second question. An an- 



Tahle 20. — Blood values for ring-necked 

 pheasant (adult males) 



[Differential counts in percent] 



swer based on some figure derived from a sta- 

 tistical analysis takes into account the variability 

 within and between groups and thus could be dif- 

 ferent, depending on the scope of the population 

 under study. In the Atomic Energy studies on 

 two inbred lines of Single Comb White Leghorn 

 chickens, birds 6 weeks and 12 weeks of age in 

 groups of 15 to 25 birds were used, but in laying 

 hens the numbers were reduced for various rea- 

 sons to 7 to 12 birds in a group. Working with 

 these data led to the conclusion that 10 was the 

 minimum desirable number in an experimental 

 group and that 15 was preferable. 



This is only a rough estimate derived from 

 studies by Lucas and Denington on whole body 

 X-ray irradiation. In this same study they 

 found (unpublished data) that if the coefficient 

 of variability for the 95-percent fiducial interval 

 were to be held to some fixed value, for ex- 

 ample, ±10 percent of the group mean, then 

 the following number of birds would be needed 

 for a study on each of the components of the 

 blood : 



Number 

 of chickens 

 Blood component: in a group 

 Red blood cells/mm.=' 10 



Hemoglobin, gms./lOO ml 12 



Hematocrit in percent 8 



Buffy coat in percent 33 



Thrombocytes/nim.' 140 



White blood cells/mm.^ 106 



Lymphocytes/mm.^ 130 



Monocytes/mm.^ 352 



Heterophils/mm.^ 240 



Eosinophils/mm.^ 3615 



Basophils/mm.^ 551 



Table 21. — Blood values for laboratory pigeon 



[Differential counts in percent] 



1 100 cells counted from each bird. 



' 100 cells counted from each bird. 



217 



