100 



MISSOURI BOTANICAL GARDEN. 



would have involved a considerable amount of labor. The 

 plants which furnished the material were of varying ages. It 

 is not safe to refer any differences observed in the constants 

 from year to year to any conditions under which the plants 

 flowered, for the proportion of the flowers which each indi- 

 vidual contributed may vary from year to year. The data 

 are given in the form of correlation tables showing the rela- 

 tionship between sepals and petals, sepals and stamens, and 

 petals and stamens in Tables I to VIII. 



Considering first the variabilities, we note that the distri- 

 butions arc very irregular, falling wide of the symmetrical 

 Gaussian distributions which some biologists of great repu- 

 tation still hold to represent the variability of all organisms. 

 But it will hardly be profitable to discuss skewness or other 

 statistical characteristics of these distributions. For such 

 work larger series grown under less abnormal conditions 

 should be used. 



TABLE A. 



VARIATION CONSTANTS FOR LAGERSTROEMIA. 



Series 



Average and 

 Probable Error. 



Standard Deviation 



and Probable 



Error. 



Coefficient of 

 Variation. 



Sepals, 1903 



Sepals, 1904 



Sepals, 1907 



Petals, 1903 



Petals, 1904 



Petals, 1907 



Stamens, 1903 . . . 

 Stamens, 1904. . . 

 Stamens, 1907 . . . 



5.879 ±.029 

 5.969 ±.008 

 5.758 ±.011 



5.879 ±.029 

 6.024 ±.006 

 5.718±.012 



4.020 ±.150 

 5.996 ±.007 

 3.686 ±.052 



.433 ±.021 

 .338 ± .006 

 .441 ±.008 



.433 ±.021 

 .260 ±.005 

 .488 ±.009 



2.214 ±.106 



.273 ±.005 



2.093 ±.037 



7.362 

 5.764 

 7.657 



7.362 

 4.316 

 8.525 



55.061 



4.557 



56.719 



The means and standard deviations, with their probable 

 errors, and the coefficients of variation, are given in Table A 

 for the three characters studied. 



Comparing the constants for the three years, we see that 

 the means for sepals and petals agree fairly well, though dif- 

 fering by several times their probable errors. The means 

 ,of the large stamens, however, are widely different in the 



