114 MISSOURI BOTANICAL GARDEN. 



There are sources of danger in interpreting these coefficients 

 of correlation biologically. We are dealing here with statis- 

 tical constants which measure precisely the degree of simi- 

 larity of the associated variables under consideration, but 

 give no intimation whatever as to the cause of this simi- 

 larity. The physiologist has long used the term correlation, 

 and it must be borne in mind that the word has a somewhat 

 different meaning in the terminology of quantitative biology. 

 All that the coefficient of correlation does is to describe, not 

 interpret, degrees of interdependence. In the present in- 

 stance we are deahng with perennial plants, and it is not at 

 all unreasonable to think that the age of the individuals 

 may make considerable difference in the magnitude of both 

 the length of the flowering-stalk and the number of flowers 

 per inflorescence. In this case the correlation would be in- 

 creased by an amount depending upon the age heterogeneity 

 of the collections. There seems to be, however, no way of 

 avoiding this difficulty in material of this kind gathered in 

 nature. Comparison with plants of annual habit of growth 

 might, perhaps, throw some light upon this question. 



In conclusion, the data presented show that there is a 

 considerable degree of interdependence between the length 

 of the flowering-stalk of these monocotyledons and the num- 

 ber of flowers which they produce. It is not at all surpris- 

 ing to find this relationship, and refined statistical methods 

 would not have been required to demonstrate its existence 

 in this particular case. To those who urge that it is pedantic 

 to devote laborious researches to the demonstration of what 

 is to be a priori expected, the reply is two-fold. First, in 

 science, assumptions, however plausible on the surf ace, should 

 never be accepted without actual demonstration on adequate 

 material. The need of careful regard to this dictum is found 

 in the group of problems we are considering here. In un- 

 published data somewhat similar to these, for other species, 

 I find results not at all in agreement with those set forth in 

 the present paper, although there is a priori quite as much 

 reason for expecting high positive correlations as in the spe- 

 cies here considered. Second, we should not be content to 

 merely demonstrate the existence of a relationship; we 



