88 INTRODUCTION 



Cucurbitaceae (Trianosperma), of which the countless flowers are odourless, greenish, 

 and quite inconspicuous, while most of them are hidden under the foliage. In spite 

 of this these plants appear to have a quite remarkable power of attracting bees. 

 There is a constant humming and buzzing around them the whole day long ; Apis 

 mellifica is the chief visitor, and there are also two Meliponae.' This phenomenon 

 might be explained either by Kerner's hypothesis or by mine. 



It is usually impossible to describe the colour of a flower, or at least the shades of 

 colour. Hermann Miiller ('Alpenblumen,' p. 502, note) calls attention to the difficulty 

 of representing objectively the colour gradations of flowers. He mentions that Koch 

 (' Synopsis,' ed. tertia, p. 499) depicts the flowers of Verbena officinalis as clear purple, 

 while they appear bluish to Miiller, and the former calls many other flowers violet 

 which the latter regards as blue. Miiller tried therefore ('Alpenblumen,' pp. 562, 

 563) to imitate as closely as possible by means of Faber's pencils the natural colours 

 of all the flowers he drew, a method which in many cases proved quite satisfactory, 

 though as a rule it attained only very partial success. 



In order to represent the hues of flowers simply and accurately, Pillsbury recom- 

 mends (Bot. Gaz., Chicago, xix, 1894) that six normal colours taken from the spectrum 

 be used as a basis, the proportions in which they are mixed to form any given hue 

 being determined (possibly with black and white) by means of Maxwell's top. Thus, 

 R. 48, V. 52, for Polygala paucifolia, indicates that the flowers of that plant contain 

 48 % red and 52 % violet. With some practice and care the determination may be 

 made with great accuracy. 



The most correct method would certainly be to determine floral colours 

 spectroscopically. This, however, is naturally only possible for uniformly coloured 

 flowers. I have frequently attempted such determinations, and did not find them 

 difficult, but the method is certainly not applicable to all flowers. In his work, 

 'Die Spektralanalyse der Bliitenfarben ' (Jahrb. wiss. Bot., xx, 1889, pp. 78-105), 

 N. J. C. Miiller has spectroscopically analj'sed the colours of sixty-five diff"erent species 

 of flowers, using a micro-spectroscope provided with a micrometer (Seybert's), 

 the illumination being efi'ected by directing the mirror of the microscope towards 

 a bright cloud. The spectra of floral leaves thus obtained show the absorption 

 bands of the pigments (expressed by shading) between the lines of Frauenhofer 

 so clearly, that this mode of representation appears thoroughly adapted for application 

 in making comparisons between different floral colours. 



Next to colour, Odour is the most important allurement for animals that visit 

 flowers, and in many cases it can scarcely be decided which of the two is the more 

 effective. It is usually the flower that is odorous, but in individual cases the smell 

 of the foliage and of the stem is obviously an attraction to insects. This purpose 

 may be served, for instance, by the strong smell of the leaves of Ruta graveolens, as 

 well as that of species of mint, lavender, marjoram, &c. 



In some cases there is without doubt a relationship of vmtual exclusion behveen 

 colour and odour in flowers, i.e. there are numerous flowers with very gaudy, 

 conspicuous colours, which are odourless, while on the other hand many very 

 inconspicuous flowers possess a strong odour. To the former group belong, e.g., 

 the conspicuously coloured poppies (Papaver Rhoeas, dubium, somniferum, Argemone), 

 Glaucium, Chelidonium, Adonis aestivalis and autumnaUs, Camellia japonica. Azalea 



