CERTAIN LAWS OF VARIATION. 207 



animals should be very great indeed. Upon this point 

 there is, unfortunately, very little evidence, though the 

 general opinion of embryologists supports the view. 

 However Fischel * has made several different measure- 

 ments on 104 embryos of the duck, and he classified his 

 results according to the number of vertebrae they con- 

 tained. He found that the variability was very great, 

 and also that it diminished considerably with growth, 

 but the values are too irregular for brief quotation. 



As the variability of guinea-pigs becomes nearly 

 halved during post-embryonic development, it fol- 

 lows that any retardation or acceleration of growth 

 produced in the earliest stages must also become nearly 

 halved, or perhaps still more diminished. A certain 

 amount of variation must have been produced in the 

 growth of the animals by variation in their environ- 

 mental conditions during development, though prob- 

 ably, as these conditions appear to have been kept as 

 constant and as favourable as possible, this was not 

 very large. One may perhaps assume, therefore, that 

 if absolutely constant environmental conditions had 

 been maintained, the variability, when growth had 

 ceased, would not have been reduced to very much less 

 than half its original amount. Now by the end of the 

 first year the rate of growth had diminished to about a 

 fiftieth its amount at birth, and hence, in order to pro- 

 duce equal permanent effects upon the growth, the 

 period of exposure of an animal to a changed environ- 

 mental condition shortly after birth would probably 

 need to be only about a twenty-fifth part of the time 

 required a year later. Again, to revert to the data ob- 

 *Morph. Jahrb., xxiv. p. 369. 



