262 PROCEEDINGS OF THE AMERICAN ACADEMY 



Semina humefacta plerumque (ut Collomice) mucilaginosa, in omnibu9 

 oppositifoliis palraatifidis nee spirillifera. Herbse, paucse suffrutices, 

 Labitu, varire. — Gilia et Navarretia, Ruiz & Pav., Benth. in DC. 



Thus regarded, Gilia is certainly a polymorphous as well as a large 

 genus; but definite characters are vainly sought for dividing it and for 

 keeping Navarretia separate. The most natural separation would seem 

 to be into three genera, characterized mainly by the foliage : — namely, 

 1. Gilia, with alternate and pinnately cut or divided leaves ; 2. Lepto- 

 dactylon, frutescent plants, with nearly the corolla of Phlox, and alter- 

 nate palmately parted leaves ; and 3. Leptosiphon, annuals, with oppo- 

 site and palmately divided (or entire) leaves. And to this the seeds in 

 some sort answer, — those of the first being mostly mucilaginous and 

 spirilliferous, as in Collomia, of the second (always?) unaltered in 

 water, as in Phlox, one section of which it externally resembles in fo- 

 liage ; -of the third, more or less mucilaginous, but destitute of spiricles ; 

 which is paralleled by the one Collomia, C. gracilis, that tends to have 

 opposite leaves, — points worth noticing by those who accept the doc- 

 trine of the derivation of species. -But Nuttall's Siphonella and a new 

 opposite-leaved Leptodactylon nearly efface the distinctions between the 

 latter and Leptosiphon ; some species of the opposite and palmate 

 series have the upper leaves prevailingly alternate ; one of the alter- 

 nate-leaved series has trisected leaves seemingly of the palmate sort; 

 and a few scattered species of the same series have seeds which pro- 

 duce neither simple mucilaginous tubes nor spiricles when wet. Those 

 of G. {Ipomopsis) coronopifolia differ in. this way from those of the 

 nearly related G. aggregata. Similarly G. (Linanthns) dicholoma has 

 seeds with a loose arilliform external coat, under which are apparently 

 no mucilage cells or tubuli, while these abound under the closer coat in 

 the nearly related G. Bigelovii, as in the other species of that series. 

 It is obviously impracticable, therefore, to restore any of those, at first 

 apparently well-marked genera which Mr. Bentham proposed, and 

 afterwards merged in Gilia. To complete our view of the genus I 

 have included the few South American species. 



Series I. Palmati- seu Oppositifolice, nempe foliis sessilibus palma- 

 tisectis (segmentis angustis integerrimis) in perpaucis integris, oppositis 

 vel summis ramealibusque quandoque alternis, in Leptodactylis pleris 

 alternis. Semina humefacta tegumento sogpius mucilaginoso sed nun- 

 quam spirillifero. 



