INTRODUCTION AND METHODS. 5 



completeness. The observational method has necessarily been continued 

 in connection with normal pollination and the life-history of the flower, 

 but it has been made as quantitative as possible and has frequently been 

 supplemented by experiment. The experimental methods have been made 

 as simple and direct as possible and have been applied chiefly to attraction, 

 behavior at the flower, and competition between flowers in this first series. 

 Most of them have been subjected to repeated check and the results are 

 thought to be fairly conclusive for the region and the time concerned. 

 The conditions have differed materially from those obtaining in practically 

 all other studies in this field, in that both flowers and insects were in the 

 natural relations that have existed for a long period. While gardens yield 

 behavior results of as much interest as those of natural habitats, it is obvious 

 that their essentially exotic nature renders them of little value in questions 

 of adaptation and evolution. This difference is also to be taken into account 

 in other respects; for example, artificial flowers appear to be much more 

 readily visited in gardens than in nature, as would be expected from the 

 difference in the habits of the insects. 



Moreover, it is thought that the experimental investigation of a native 

 insect fauna in the midst of its natural floral environment, and the con- 

 verse, furnishes a norm to which can be referred other studies that are 

 artificial in some degree. Some such standard appears indispensable, 

 since it is evident that many of the contradictions and discrepancies in the 

 results of various investigators are to be explained by differences in con- 

 ditions and setting rather than by faults of method or observation. This 

 is certainly true of many of the points at issue between Plateau and his 

 critics. The effects of time, place, weather, grouping, etc., are often 

 decisive, as shown in the next chapter, and they need always to be checked 

 by actual trial, or, much better, eliminated by simultaneous observations 

 in contiguous areas in so far as possible. The time of day, week, or season 

 not only has an effect due to lapsed time, but also one especially of differ- 

 ence in sun, cloud, temperature, wind, condition of flowers, habit of insects, 

 etc. Even the weather of the previous day may have a profound effect, 

 if it has been rainy, unusually cold or warm, or windy. Differences of 

 location and particularly region usually produce decided effects, owing to 

 changes of conditions as well as of flower and insect populations. In fact, 

 marked differences of behavior have been noted in spots a yard apart 

 where no differences of sunlight, temperature, or wind were demonstrable 

 at the time, but where the behavior of the insects had been determined by 

 earlier shade, exposure, by nearness to their nests, etc. The kinds, number, 

 and grouping of the species and individual flowers naturally have a pro- 

 nounced effect, as do also the time of blooming, the position in the flowering 

 period, the nectar flow, and the rate of nectar production. It is equally 

 evident that the composition of the insect fauna as to orders and species, 

 the number of individuals, the sexes, social habits, age, etc., will greatly 

 affect the results. Finally, it has been found that the position and nearness 

 of the observer, as well as his clothing (cf. Lovell, 1914:407), produce 

 effects to be taken into account, while an increase in the number of observers 

 in the same spot may completely change the response. Still other factors 

 enter into the problem, thus completing the certainty that researches 



