in I STUDIES IN INDIAN SUGARCANES 



canes of different lengths, the process of averaging largely rules them out. 

 And this question of maxima is coniplicated by the not infrequent appearance 

 of second, or evei' tliird and fourth maxima, as if there was a more or less 

 defin.ed j)eri()dicity in the growth in length of the j)arts. Iii. the general summa- 

 tion, these secon.dary maxima are als.o mostly luled out. 'J'his is a serious 

 drawback, but. on the other hand, local, excessive variations, in. individual 

 canes, which may be due to many causes, are also ruled out, which is not 

 such a disadvantage. It is thus quite possible to have a flat summation 

 curve, in. which the individual can.es show several maxima and great 

 individual variations from joint to joint. But these maxima should, as far 

 as possible, be represented in the ideal cane, and their inclusion is on.e of the 

 chief difficulties in. our problem, 



The following method, although somewhat arbitrary, has been, design.ed 

 to give each maximum when possible its value, it being inevitable that all of 

 them are much less marked m the summation series than, in the in.dividual 

 canes. In the table, each cane ii; studied as regards its maxima, and these 

 are iii.dicated by dots. Lines are then, drawn between the maxima in. adjoin.in.g 

 canes, and dotted, hnes if the canes are not next to one another. By this system 

 the general trend of maxima in the whole series can be seen at a glance. In. 

 the table given as an example, there is a certain amount of periodicity of growth, 

 and three separate maxima are often present. The first six and tlie last ten. 

 can.es of the table agree very closely in their periodicity. Attention, is now 

 paid to the summation, series at the bottom representing the averages of the 

 column.s. If this series shows maxinia in the right places, as judged by the 

 rows of dots and lines, and. this is not infrequently the case, the figures are 

 left intact, providing that there is no sudden jum_p to the last eight immature 

 members because of the removal of unrepresentative columns. If the maxima 

 are not shown, where they would be expected, or if there is an. unjustifiable 

 break in the curve just before the last eight, as is often the case, arbitrary 

 alteration.s are made. But this is carefully safeguarded by making the 

 alterations chiefly in. the non-representative series between the arrows. The 

 amount of actual change is, however, very little, as will be seen from the 

 typical, but rather simple, case shown, in. the table. Two maxima are duly 

 reflected in the average series (the second maximum representing the second 

 and third of the individual canes merged into on.e), and it is merely necessary 

 to alter a few numbers, so that there is a less sudden fall, ii\ the figures, to the 

 last eight, than is justified fr(uu a study of the whole twenty in this 

 region. 



