MEMOIRS FROM THE DEPARTMENT OF BOTANY OF COLUMBIA UNIVERSITY 11 
The parapetalous stamens must be regarded as abnormal supernumerary parts, very 
common in the Rosackar and rather characteristic. Any theory as to their origin can 
be only a matter of speculation. I have seen no explanation given. As stated above, 
many authors have regarded the bractlets as representing the united stipules of two ad- 
jacent sepals. This suggested to me that the parapetalous stamens might have the same 
relation to the antipetalous stamens as the bractlets have to the sepals, only with the 
difference that they have remained free. Their position, viz., one on each side and a 
little outside of the antipetalous stamens, is the one to be expected of such modified 
stipules. Acting on this suggestion, I traced the fibro-vascular bundles of those stamens 
in Horkelia sericoleuca, Pickeringii and argyrocoma and found that they united with those of 
the nearest antisepalous stamens (these species have parasepalous instead of parapetalous 
stamens) before they reached the bottom of the calyx, as shown in plate 1, fig. 8, us. 
This fact would suggest that they represent in a certain way the stipules of the antisepa- 
lous stamens, The fact that they develop before the antisepalous stamens is also par- 
allel to the earlier development of stipules. 
In several species of Potentilla with 30 stamens as for instance, P. recta and sul- 
phurea (fig. 14), the additional series of 10 stamens (ps) have exactly the same position 
relative to the antisepalous stamens (as) as the outer series of 10 parapetalous (pp) 
stamens has to the antipetalous stamens (ap). That would mean that not only the anti- 
petalous but also the antisepalous stamens had stipules. There 1s, however, another 
explanation, perhaps neither better nor worse, viz., that the antipetalous stamen and its 
two neighboring parapetalous stamens represent a trifoliolate leaf. The Horkeliae with 
20 stamens have apparently parasepalous instead of parapetalous stamens (fig. 8). 
It must be remembered that the presence of supernumerary stamens is not uncom- 
mon, Such a condition is often found in the antipetalous and antisepalous stamens, so 
that two stamens instead of one are found opposite one or more petals or sepals. The 
parapetalous (and parasepalous, if present) are seldom duplicated in flowers with less 
than 30 stamens. There are very few species of Potenti/la that have a greater number, 
and in such eases it is nearly impossible to make out the arrangement. Of the North 
American species most have about 20, some as many as 30, and one Mexican species 
about 40 stamens. 
Many species of the tribe have fewer than 20 stamens. In such cases it is generally 
the parapetalous series that first disappears. The number then becomes 10, as com- 
monly in Potentilla rivalis, P. leucocarpa, P. biennis (Fig. 10) and Horkelia (Ivesia) 
pygmaca. If further reduction occurs, which is even more common, the antipetalous 
stamens are also lacking and the androecium is represented by the 5 antisepalous 
