288 TRANS. ST. I.(JUIS ACAD. SCIENCE. 



In restorin<^ O. recurva* which has long been lost as a synonym 

 of O. stricta, I found that It has an extremely close representative 

 in the Oregon plant that forms the subject of Mr. Eliot's paper. 

 This differs mainly in its smaller size and more deeply notched 

 leaflets, destitute of a dark marginal line found on those of O. re- 

 curva. If the two plants occurred together, or in adjacent regions,^ 

 one might question the propriety of regarding them as more than 

 varieties of a single species ; but the complete isolation of O. Suks- 

 dorjii^ which is said to be very abundant about Portland, Oregon^ 

 but appears to be very local in its occurrence, weighs very heavily 

 against this union. Both are peculiarly American species, and 

 with'>ut much doubt both are descended from an ancestral form 

 essentially like O. stricta. 



The trimorphism of O. Suksdorfii, which was clearly shown by 

 herbarium specimens, is demonstrated to entire satisfaction by the 

 measurements made by Mr. Eliot. The diagram made by plotting 

 his measurements on coordinate paper, and connecting the points 

 so obtained, shows at a glance the peculiarities of this trimor- 

 phism, though it is probable than in examining a thousand or 

 more plants the relative number of long, short, and mid-styled 

 flowers would have been more nearly equal ; and the last meas- 

 urements on the sheet — from a very short mid-styled flower — 

 carries the curves for this form very much below their normal 

 level. 



In the diagram of a trimorphic species without variability, one 

 would expect to find the lines i, 4, and 7 coincident at the top of 

 the paper ; 3, 6, and 9, at the bottom ; and 2, 5, and 8 superposed 

 acro>s the middle of the sheet With a small amount of varia- 

 bility in the flowers, each set of three curves should interlace and 

 remain quite distinct from the other sets. This is, indeed, very 

 well shown in the upper set (1,4, and 7), and, as Mr. Eliot has 

 observed, would be still more evident if a like number of each 

 kir.d of flowers had been measured, so as to prolong all of the 

 curves across the sheet. The same relations are observable be- 

 tween 3, 6, and 9. 



On seeing the curves for the first time, I was immediately im- 

 pressed by the fact that only two (5 and 8) of the other set lie 



* Memoirs Boston Soc. Nat. Hist., vol. iv. No. 4. ; Bet. Gazette, xii. 166. 



