Whether this explanation can be considered satisfactory must remain at present doubtful.
There is another remarkable difference between the forms of several heterostyled species, namely in the anthers of the short-styled flowers, which contain the larger pollen-grains, being longer than those of the long-styled flowers. This is the case with Hottonia palustris in the ratio of 100 to 83. With Limnanthemum Indicum the ratio is as 100 to 70. With the allied Menyanthes the anthers of the short-styled form are a little and with Villarsia conspicuously larger than those of the long-styled. With Pulmonaria angustifolia they vary much in size, but from an average of seven measurements of each kind the ratio is as 100 to 91. In six genera of the Rubiaceae there is a similar difference, either slightly or well marked. Lastly, in the trimorphic Pontederia the ratio is 100 to 88; the anthers from the longest stamens in the short-styled form being compared with those from the shortest stamens in the long-styled form. On the other hand, there is a similar and well-marked difference in the length of the stamens in the two forms of Forsythia suspensa and of Linum flavum; but in these two cases the anthers of the short-styled flowers are shorter than those of the long-styled. The relative size of the anthers was not particularly attended to in the two forms of the other heterostyled plants, but I believe that they are generally equal, as is certainly the case with those of the common primrose and cowslip.
The pistil differs in length in the two forms of every heterostyled plant, and although a similar difference is very general with the stamens, yet in the two forms of Linum grandiflorum and of Cordia they are equal. There can hardly be a doubt that the relative length of these organs is an adaptation for the safe transportal by insects of the pollen from the one form to the other. The exceptional cases in which these organs do not stand exactly on a level in the two forms may probably be explained by the manner in which the flowers are visited. With most of the species, if there is any difference in the size of the stigma of the two forms, that of the long-styled, whatever its shape may be, is larger than that of the short-styled. But here again there are some exceptions to the rule, for in the short-styled form of Leucosmia Burnettiana the stigmas are longer and much narrower than those of the long-styled; the ratio between the lengths of the stigmas in the two forms being 100 to 60. In the three Rubiaceous genera, Faramea, Houstonia and Oldenlandia, the stigmas of the short- styled form are likewise somewhat longer and narrower; and in the three forms of Oxalis sensitiva the difference is strongly marked, for if the length of the two stigmas of the long-styled pistil be taken as 100, it will be represented in the mid- and short-styled forms by the numbers 141 and 164. As in all these cases the stigmas of the short-styled pistil are seated low down within a more or less tubular corolla, it is probable that they are better fitted by being long and narrow for brushing the pollen off the inserted proboscis of an insect.
With many heterostyled plants the stigma differs in roughness in the two forms, and when this is the case there is no known exception to the rule that the papillae on the stigma of the long-styled form are longer and often thicker than those on that of the short-styled. For instance, the papillae on the long-styled stigma of Hottonia palustris are more than twice the length of those in the other form. This holds good even in the case of Houstonia coerulea, in which the stigmas are much shorter and stouter in the long-styled than in the short-styled form, for the papillae on the former compared with those on the latter are as 100 to 58 in length. The length of the pistil in the long-styled form of Linum grandiflorum varies much, and the stigmatic papillae vary in a corresponding manner. From this fact I inferred at first that in all cases the difference in length between the stigmatic papillae in the two forms was one merely of correlated growth; but this can hardly be the true or general explanation, as the shorter stigmas of the long-styled form of Houstonia have the longer papillae.