110 



THE BOTANICAL MAGAZINE. 



[Vol. X.KII. No. 257. 



Nine days later I made the fourth visit. It was late in 

 the season, and the flowers, having no j'oung bud, were seen 

 onh^ in the lowest part of the area. Following results were 

 obtained : 



Fourth collection. Aug. 30. 



In the lot from the valley, general feature of the frequency- 

 poh'gone (Fig. 6) was somewhat alike to that of the former 

 collection there (Lot No. -i), except the disappearance of higher 

 variates and the accessory mode on 1-i instead of 13. The 

 mean and C was equally less than those of the lot No. •A, but 

 nearh^ equal to those of the lot No. 5. 



In that of the slope the distribution of variates was very 

 different (Fig. 7). No higher variate was seen than that with 

 20 rays. The mode on 16, which was hitherto always the 

 case, has entirely disappeared, and the polygon, with the 

 highest mode on 12 and with no sinus on 15, seemed like a 

 obliquely truncated column extending from 12 to 16. The 

 variates on other classes were very small in number, and the 

 S. D. was the least of all the lots. The mean was also ex- 

 tremely small. 



If we compare now all these sevea lots with each other, we 

 can find some interesting facts among them. Firstly, as to the 

 mean, it reached its maximum value in the second collection, 

 since then it decreased continuously. In the first collection, 

 however, the mean was less than that of the second, though 

 the duration of annual growth is longer than the latter. In 

 the multicipital compositae hitherto studied, it was claimed, 

 that " those heads which blossomed first had a prevailing 

 larger number of parts than those which appeared later in the 



