198 INTRODUCTION 



From this table it appears, for example, that Class E attracts allotropous 

 Diptera more than the other classes. The proportion (20 %) of visits of Diptera to 

 this class of flowers is higher than the proportion (8-6 %) of visits of insects in 

 general, from which it may be concluded that Class E possesses means of allurement 

 by which allotropous flies are specially attracted. By comparing the table relating to 

 a particular insect group with that for insects in general, it is therefore possible 

 to determine for a given month and district the flower classes which the insect 

 group in question prefers or avoids. 



MacLeod has worked through on this method Miiller's observations in the Alps 

 and Loew's observations in the botanic gardens at Berlin, and has drawn up ten 

 diff'erent series according to the month and place of observation, so that the choice 

 of flower of each insect group could be determined ten times. 



The clearness of MacLeod's method is enhanced by the fact that he gave 

 a graphic representation of his results. For each month he erected upon a 

 horizontal line at equal distances from one another seven ordinates 100 mm. 

 long, corresponding to the flower classes, and then measured off" on each ordinate 

 a length representing the proportion in which the class in question was visited by 

 insects in general. By connecting these points with one another, he obtained a 

 broken line which he described as the general insect line. In similar fashion he then 

 drew special lines for the individual insect groups, so that the choice of flowers made 

 by the several groups could be seen at a glance. Wherever a special line runs 

 above the general insect line, the corresponding insects must have a preference for 

 the particular flower class, and, conversely, the deeper a special line sinks below the 

 general insect line, the greater is the repugnance manifested by the particular insect 

 group towards the flower class indicated. To be of any value this graphic method 

 must give constant results, and must therefore yield the same result for the same 

 insect group and the same flower class, in each of the ten series of observations 

 collated by MacLeod. Such uniformity obtains for the following cases : 



1. Beetles show a constant preference for An and E, while C, H, and L are the 

 most repugnant to them. 



2. Allotropous Diptera prefer E, and always reject H. 



3. Hemitropous Diptera consistently prefer EC, and reject H. 



4. Short-tongued Bees always avoid H. 



5. Long-tongued Bees avoid E and S, and consistently prefer H. 



6. Lepidoptera consistently prefer L, and reject E. 



Although in other cases there was no agreement between the ten series of 

 observations collated by MacLeod, it was demonstrated that the groups of flowers 

 and insects are not homogeneous. In those cases where, on Muller's theory, 

 strong preference or the opposite could be inferred of certain visitors in regard to 

 certain flower classes, reasoning from the structure of the insects and of the flowers 

 they visit, MacLeod's results were constant. 



By this graphic method relations can be recognized which, although theoreti- 

 cally probable, were not deduced by the older ways of dealing with statistics, e.g. 

 the preference of Lepidoptera for lepidopterid flowers comes out very clearly. 

 MacLeod's method would furnish still more reliable results by taking smaller 



