196 INTRODUCTION 



The value of this method, continues Loew, lies also in the fact that it is not 

 a purely statistical one, furnishing only averages, but specifically oecological, by 

 which the behaviour of individual insects with regard to flowers can be quite as well 

 established as the behaviour of a complete group. The method has afforded 

 valuable proof of the general correctness of Miiller's rule of colour-preference, 

 i. e. that insects which are more perfectly adapted to pollination generally affect 

 dark colours (blue, red, violet), while insects which are less specialized prefer light 

 colours (white, yellow). And, lastly, it is proved that insects choose flowers in the 

 series as they would be expected to from the structure and length of their suctorial 

 apparatus, and the nature of their general bodily equipment, i. e. Ihose pollinators and 

 flowers which theoretically appear to be adapted to one another, are the very ones which 

 actually exert the most marked mutual attraction. This proposition of Loew's 

 (' Beitrage,' p. 15) was previously postulated by Muller, but was only proved by 

 later statistics. 



The two following conclusions embody the general results of the statistical 

 investigations made by myself in accordance with this method (' Bliitenbesucher,' 

 II, p. 9). 



1. The more specialized a flower i.e. the more complex its structural arrange- 

 ments and the more deeply seated its ?iectar the less are its insect visitors indiscriminately 

 drawn from the entire insect fauna of a district, and the more do they belong to one or 

 several similar species adapted to pollinatiofi. 



2. The flatter and more superficial the position of the nectar, the more varied are 

 the visitors in difl^erent regions, and the more are they indiscriminately drawn from the 

 entire insect fauna of the region in question. 



MacLeod (' De bevruchting der bloemen door de insekten,' Verhandlingen van 

 het eerste Nederlandsch Natuur- en Geneeskundig Congress, gehuiden te Amsterdam, 

 op 30 Sep.-i Oct. 1887, Amsterdam, 1888) has tested Miiller's method from a 

 new point of view. This investigator (op. cit., pp. 83-90) proceeds from the idea 

 that always and everywhere the degree to which any particular group of insects 

 visits a particular class of flowers depends upon three factors, i.e. i. On the Choice 

 of Flowers by insects, or in other words, on the preference of insects for certain 

 flowers. 2. On the Composition of the Flora, i. e. on the proportions in which the 

 various classes of flowers are represented in a district. 3. On the Time of Year, by 

 which quite different species of flowers are offered to visitors during different months. 



The first of these three factors, according to MacLeod, is constant for similar 

 flowers and similar insects, which must be determined by statistics. The two other 

 factors are varying quantities which must be eliminated before we can attain to 

 a constant result. 



The influence of the Time of Year can easily be eliminated by separating from 

 one another the observations for various months and regions, and considering them 

 individually. MacLeod divides the summer half-year (in central Europe from the 

 first of April to the first of October) into intervals of thirty days, and places the 

 observations in so many series according to their date. Each result is therefore 

 repeated as often as there are monthly series, so that the degree of reliability of each 

 conclusion can be estimated. It may be assumed that the plants in flower during 

 a particular period of thirty days remain much the same. 



