87 

 anther lies 8o close to its opposite neiglibor and they dehisce 

 so widely on their opposed faces that both together form one 

 pollen reservoir." They are closed together by the hairs on 

 the edges of the anthers. The filaments are rigid. At their 

 bases the anterior pair is set clo.se together and provided with 

 sharp points on their inner sides but a short distance below 

 the anthers they are smooth and wide enough apart to admit 

 the head of an insect. When the bee's tongue is pushed in, 

 the filaments are forced apart, the anther cells are torn and 

 the poUen falls on the insect. 



Yellow Rattle produces two kinds of flowers, one con- 

 spicuous and unable to be self-fertilized, the other conspicu- 

 ous and regularly self- fertilized. 



For a general synopsis of the Lahiateae, a pa- 

 per by Robertson and Hermann Mueller's Ferti- 

 lization of Flowers should be consulted Robertson says: 



"In the case of the UmbelUferae, which have very uni- 

 form flowers, I have been able to show that the preponder- 

 ance of the. bees, lower Hymenoptera or Diptera was deter- 

 mined mainly by the time of blooming, and the peculiarities 

 in the lists of visitors were best shown by a table arranging 

 species in the order of blooming A similar arrangement of 

 the Labiatae does not give very important results, since the 

 differences in the lists are mainly due to differences in struc- 

 ture, and the time of blooming has little influence. In the 

 case of the Pycnanthemums, there seems to be a plain case of 

 correlation between the form and arrangement of the flowers 

 and the time of flight of the lower Hymenoptera, which are 

 very important visitors. If the flowers bloomed in April the 

 lower Hymenoptera would be almost entirely wanting, and 

 the list would consist mainly of bees and flies. 



Of the twenty-three species considered in the present 



