72 MORPHOLOGY OF MALLOW-WORTS. [BOOK i. 



We thus have ten radiant series, opposed in pairs to the petals, 

 and supported upon a common base, which is frequently cut 

 into five corresponding lobes, more or less marked. At a 

 little later period, each of these tubercles, continuing to grow 

 more at the sides than in the median line, is itself divided 

 into two, and we find that four parallel series become substi- 

 tuted for the two before each petal, and the total number is 

 a second time doubled. The same occurs in those flowers 

 which have very numerous stamens; but there is a slight 

 difference in those in which they exist in less numbers. 

 When either fewer concentric rows are formed, or each of 

 these rows stops at that period at which the pairs are simple 

 and not doubled, or within the first pairs, a single tubercle 

 only is formed; this is slightly lateral and oblique; then 

 another still more internal and on the opposite side, so that 

 within the first pair we find only isolated tubercles, sent off 

 alternately, first from one side, then from the other, in a zig- 

 zag direction. In all cases, there are invariably five systems 

 of stamens opposite to the petals. 



During these changes, the small common tube, to which all 

 these organs are attached, continues to elongate, raising these 

 concentric formations so as to produce a system of stages 

 arranged one above the other ; and, although they enlarge at 

 the same time, they do not do so in the same proportion. 

 The organs which enlarge do not then find sufficient room to 

 lie side by side in regular and concentric circles ; they become 

 rather confusedly mixed, and the original symmetry becomes 

 less and less apparent. When they have arrived at a certain 

 degree of development, each of the tubercles shrinks up at 

 the base into a minute filament which becomes more and 

 more elongated. Each also becomes marked by a median 

 furrow, and buried within two cells which subsequently fuse 

 into a single one. In short, these are so many reniform, 

 unilocular anthers, which tend more and more to assume 

 their definite form. 



In several species M. Duchartre has observed an ulterior 

 change, from which a new increase in the number of stamens 

 results. Several of them are curved into a horse-shoe form, 

 and terminate by becoming divided into two by a constric- 



