The mark × indicates a partial resolution of a stamen into two, the filament bearing two anthers.

It will be seen from the above that, while P. peduncularis and P. Calabra have a tendency to vary, both in the antisepalous and antipetalous stamens, P. inclinata varies only in the antisepalous ones. In the last-mentioned species, it is remarkable how frequently a partial or complete resolution of an antisepalous stamen into two takes place.

B. Species exhibiting a tendency to reduction in the number of

P. opaca, L. (P. intermedia, Nestler). Six flowers were examined; four were normal, while the other two each wanted one antipetalous stamen.

P. Fragariastrum, Ehrh. In this species a great number of flowers have the andrœcium reduced to the 10 parapetalous stamens. Of better-developed andrœcia, I have noted the following:

antipetalous stamens is indicated by five figures, these five figures represent the number of stamens in front of the five sepals or five petals respectively, and are noted down consecutively, as they may be read off on looking round the flower.

It is to be observed that in these reductions in the number of stamens, the antipetalous evidently disappear more readily than the antisepalous ones. This is what might have been expected, as the antipetalous stamens are the younger.

Of the species falling under Type II., those which I have examined are all variable in the number of stamens; and the tendency is almost always towards a reduction in the number. In a few flowers only is a tendency to multiplication of the antipetalous stamens to be observed. I have named with some hesitation the forms occurring in the Botanic Garden; but they certainly all come under P. hirta of De Candolle's 'Prodromus.'

In both of the species which I have mentioned as exhibiting the third type of andrœcium, viz. P. fruticosa and P. rupestris, the number of stamens varies. In the festoons which the stamens form, however, five stamens occur with sufficient frequency to justify me in assuming 25 to be the typical number of stamens in each flower. In one flower of P. fruticosa I observed a stamen superposed to one of the petals. This deviation, which is evidently rare, is very interesting, as showing an approach to the other types.

P. fruticosa, L.

Number of flowers

examined.

2

2

3

Number of stamens in

the festoons.

5, 5, 5, 5, 5

5, 5, 5, 5, 4

5, 5, 4, 5, 4

It would be rash to speculate as to the probable value of the staminal arrangement in distributing the species of Potentilla into natural groups. I scarcely anticipate that it will serve as a basis for primary division of the genus, although I have little doubt that it will be found of great importance as a means of establishing, or at least limiting, minor groups. In a genus so extensive as this, my present contribution towards a knowledge of the staminal arrangements can only be viewed as a nucleus round which the results of further investigation may be aggregated. I therefore hope that any who have opportunities of examining or discovering species in the fresh state will carefully note the disposition of the stamens.

In connection with the foregoing, I would call attention to the andræcium of Nuttallia cerasiformis, which, as is known, consists of only 15 stamens, viz. 10 parapetalous and 5 antipetalous (Plate LII. Fig. 4). Such an arrangement contrasts most interestingly with the types I have described. Thus, in Nuttallia there are no antisepalous stamens; in P. fruticosa, etc., there are no antipetalous stamens; while in P. anserina, etc., there are both antisepalous and antipetalous stamens.

* In one of these five flowers a stamen occurred superposed to one of the petals, in addition to those in the festoons.

The tip of the sepal to which the festoon with seven stamens was superposed was bifid.

Adopting my theory of rosaceous andrœcia, there is no confluence of the lobes of the compound stamens in Nuttallia, i.e. there are no interstaminal lobes.

In conclusion, I must express my obligations to Mr. M'Nab, for his having kindly permitted me to make what use I required of the collection of Potentillas in the Botanic Garden, from which I obtained the greater number of the species which I examined.

EXPLANATION OF PLATE LII.

In the diagrams the sepals are shaded. The so-called petals (apices of the compound stamens) are represented in black. The stamens are represented by black spots; their relative ages (determined by observation of development or by analogy) being indicated by the size of the spots, the larger representing the older, the smaller the younger stamens. Fig. 1. Diagram representing the staminal arrangement in species of Potentilla falling under type I. This figure is reproduced from my paper on the andrœcium of Mentzelia, etc. The antipetalous stamens are represented as the most internal; but in many Potentillas (in the adult state, at least) they appear to be external to the antisepalous stamens. 2. Diagram of arrangement in species of Potentilla falling under type II. 3. Diagram of arrangement in species of Potentilla falling under type III. 4. Diagram of arrangement in Nuttallia cerasiformis. 5. Portions of young flower of Potentilla fruticosa; ec, parts of epicalyx; s, sepals ; P, petals so-called. Between the petals festoons of staminal mammillæ extend. Of the two festoons represented, one contains five, the other four stamens.

ON THE POLLEN-GRAINS OF CERTAIN RANUNCULEE AND OF LOTUS CORNICULATUS AND L. MAJOR.

BY GEORGE GULLIVER, ESQ., F.R.S.

Though the importance of the forms and structure of the elementary parts of plants has long been recognized in the grouping of the great divisions of Phanerogams, but very little aid has yet been derived from histology or microscopic research in the discrimination of near allies of this class. And no wonder, after Schleiden had concluded that a further advancement of systematic botany could hardly be expected from mere anatomy and physiology; and when, indeed, we see how much the cells and other parts of the intimate structure of the members of the same and connatural orders are alike, while the differences are comparatively few and not discoverable without many exact comparative examinations. Still, after such observations have been sufficiently extended, we shall surely acquire a valuable addition to our