1898] VARIATION IN BANUNGUL US REPENS (L.) 187 



From these figures I assume that there is only a certain amount 

 of sexual material to begin with, and if this is insufficient to form 

 the average number of stamens, the number of carpels will also be 

 below the mean. The reverse also holds good — that is, if there is a 

 superabundance of sexual matter, so that the number of stamens is 

 above the mean, the number of carpels will also lie above the mean. 

 We do not find, at least not to any extent (] '4 per cent.), that the 

 number of stamens comes up to the mean at the expense of the 

 carpel whorl, or that tlie carpels increase at the expense of the 

 stamens (2 '5 per cent.). 



COEEELATION OF VARIATION IN ALL FOUR WhOHLS COLLECTIVELY 



There are 81 possibilities of such variation (3*) — 'the mean,' 

 ' below the mean,' ' above the mean ' — in each of four whorls. 



In the combined series (1000 specimens) of these 81 possibili- 

 ties, only 35 (43 per cent.) occur. 



FIG. 5. — VARIATION IN CARPELS OF 500 SPECIMENS OF RANUNCULUS REPENS (L.) 



Second Series. 



-NUMBER OF CAKPELS- 



■> 



It may, therefore, be definitely stated that the whorls do not 

 vary independently. Particulars of the 35 are here given. 



(1. signifies the mean, 2. = below the mean, 3, = above the mean, 

 so that 1.1.1.1 means that the specimen has in the calyx, corolla, and 

 stamen and carpel whorls, each respectively, the mean number of 

 members. Similarly 1.1.1.2, for instance, signifies that calyx, 

 corolla, and stamens are at the mean, but the number of carpels is 

 below the mean, and so on.) 



