Editorial 
305 
taken in their entirety at the end of the flowering season, but 500 to 1000 
gathered at one date from one crop as a sample of the crop, and 500 to 1000 from 
a second crop at another date. But this latter method of gathering is what actually 
has been adopted in most counting on wild flowers. So far as we are aware no 
attempt has yet been made to count any character in all the flowers, throughout 
the whole season, of some given plant on a definite small area. Thus the poly- 
morphism so often noted may be wholly or in part due to heterogeneity introduced 
by the collector iiimself ; he gathers from different localities at different parts of 
the flowering seasons. A "different locality " may mean either side of an east and 
west hedge, and a " different part of the flowering season " the same day for the 
population on either side of this hedge. 
The importance of these considerations is so great that it seems absolutely 
necessary to ascertain the influence of seasonal and environmental changes on 
plants before we conclude as to their polymorphic character or as to the existence 
of " petites esj)eces " from a discussion of frequenc}'' distributions. In order to 
emphasise these points the following three papers are now published to show the 
changes which arise in the statistical constants when the gatherings are made at 
difTerent periods in the season, and fui'ther to indicate how the theory of statistics 
can be applied to test the significance or non-significance of differences in sta- 
tistical constants. 
Mr Yule shows us how the influence of year, date of gathering, and environ- 
ment on Anemone nemorom affects the statistical constants. The differences are 
quite as significant as those which in other cases have led to the suggestion of 
" petites especes." 
Mr Tower indicates a similar seasonal change in the case of the mode of 
Chrysanthemum leucantheinum. He proposes a new definition for the term 'mode,' 
but the word ' mode ' was introduced into statistics with a perfectly definite sense, 
and it seems undesirable now to alter it. "The average prevailing state of one or 
more characters of a homogeneous lot of individuals " is not a biometric definition. 
It might refer to any constant whatever of the frequency, — to the mean, the mode, 
the variability, or indeed to the whole frequency distribution itself The now 
established use of the word ' mode ' is for that value of an organ or character, at 
which the frequency of the population per unit of the character or organ is a 
maximum, — the frequency ' per unit of the character' being used, if the character 
be not discrete, in the sense of the infinitesimal calculus. The definition is clear ; it 
belongs to the theory of statistics to show us how to determine whether there is one 
or more true modes, and if there be, to settle the degree of their significance. A 
frequency distribution with more than one true mode is multi-modal, but although 
the population will then probably be heterogeneous, it is not shown to be poly- 
morphic. We take it that polymorphism means the existence at the same instant of 
the season under the same environment in a homogeneous population of two types. 
The object of the present series of papers is to indicate that much of the multi- 
modalism interpreted in the case of flowers as polymorphism is due either to 
Biometrika i 31 
