DEPARTMENT OF EXPERIMENTAL EVOLUTION. 147 



species of the Blue Mountain region of Jamaica studied by Shreve (Carnegie 

 Inst. Wash. Pub. No. 199) is 9.4 atmospheres, whereas that of the Port Hen- 

 derson strand flora, mangrove swamp, and coastal desert vegetation is 30.0 

 atmospheres. Such general averages, however, give only a fraction of the 

 information necessary. Subdivisions of each of these habitats differ in osmotic 

 pressure. Thus the salt spots of the Arizona deserts show an average of 37.1 

 atmospheres, whereas the arroyos give a mean of 13.9 atmospheres. On the 

 xerophytic Jamaican coast one locality shows an average concentration of 25 

 atmospheres, whereas another has a mean of 48. Naturally the general aver- 

 ages for a region as complex as either of those studied will depend in large meas- 

 ure upon the relative number of determinations from each of the subhabitats. 

 In detailed discussions the latter must be made the basis of comparison. 



"Furthermore, the growth form of the plant species must be taken into 

 account. Thus Arizona trees and shrubs show a mean concentration repre- 

 sented by 28.2 atmospheres as compared with 14.8 atmospheres for the winter 

 annuals. The mean value obtained for a region will therefore depend upon the 

 proportional frequency in the series of determinations based on the various 

 growth forms. The consideration of all such questions must be reserved until 

 the full data can be presented." 



METHODS. 



An important statistical study deals with the application of intra- 

 class correlation formulae to the problem of substratum heterogeneity. 

 In all the plot-culture work of agricultural institutions — ^whether in the 

 testing of the practical value of newly originated varieties of plants or 

 in fertilizer tests — the factor of the heterogeneity of the field upon which 

 the cultures are to be made must, if results of value are to be secured, 

 receive far more attention. Dr. Harris had employed a formula which 

 measures substratum heterogeneity, as expressed in actual crop yield, 

 in terms of the correlation between associated ultimate plots. It takes 

 the convenient form 



^ _ mCl)-S{v')]/m[n{n-m-p' 



'VI V2 — 2 



where p denotes the yield of an ultimate plot, Cp the yield of a com- 

 bination plot, p the mean yield of the ultimate plots, and cr^ their 

 standard deviation, n is the number of ultimate plots in each of the 

 m combination plots, and S indicates summation, for either the ulti- 

 mate or combination plots of the field. In this sample formula n is 

 supposed to be constant, but calculation with n variable is only shghtly 

 more difficult. 



Application of this formula to the best series of data available in 

 the agricultural literature shows that, in practically every field dealt 

 with, substratum heterogeneity might have had a marked influence 

 upon the results of a comparative test. 



A paper on spurious values of intra-class correlation coefficients 

 arising from disorderly differentiation within the classes has also been 

 pubHshed by Dr. Harris. 



Considerable attention has been devoted to the problem of main- 

 taining white rats so that they should not be a nuisance in the building 



