178 POLEMONIACEAE 



races." Continued field study over wide areas for several decades gives, however, a contrary im- 

 pression. Such developments should not be regarded as odious because defeating formal system- 

 atic methods. The numerous small races of Gilia tenuiflora in its broad sense seem to represent 

 symmetrical fragments of a large pattern. If so, aggregates of this sort would seem to offer 

 a highly tempting field of exploration to the systematic botanist who can evolve a suitable method 

 of experiment and study. Meanwhile, it here seems desirable to describe the variations of various 

 organs and to describe under varietal names such extreme phases of subspecific races as it is pos- 

 sible to segregate geographically and to some extent morphologically, that is, limiting description 

 to the most characteristic collections. 



Leaf size varies relatively little, but leaf segmentation varies greatly — especially in depth 

 of segmentation and in shape and spacing of the segments: (a) In pinnately divided leaves with 

 long segments, the segments may be acute, discrete, remote or crowded, linear and entire or mostly 

 sparingly toothed, or again the leaves may be saliently or laciniately pinnate-toothed with the 

 rachis broad, (b) In pinnately divided or lobed leaves the short segments may be obtuse, approxi- 

 mate or crowded, merely toothed or again pinnately parted or divided. Two diverse leaf forms 

 may often be found in one collection. For example, peetinately pinnatifid leaves and entire leaves 

 may be found in one colony, as in Pacoima Caiion, San Gabriel Mts., Peirson 385. Simply pinnate 

 and twice pinnate leaves occur in such material as Mission Soledad, Salinas Valley, Jepson 8434a. 



The flowers vary greatly also, not merely in one collection but even on a single individual — in 

 size of corolla and especially as to relative length of corolla-tube and throat. Two collections may 

 seem to belong to one close taxonomic unit, save that in some one flower factor they may differ 

 markedly, as in relative length of corolla-tube and corolla-throat (San Jacinto Catiou, San Jacinto 

 Mts., Jepson 1284, the corolla-tube not exceeding calyx; Mt. Wilson, Peirson 155, the corolla-tube 

 1% times as long as the calyx). Conversely two collections may agree in flowers and in other 

 features but differ in leaves. These two factors, flower-shape and leaf-shape, do not, therefore, 

 necessarily vary together, nor do they characterize forms with geographic meaning. 



Certain broad features of GiUa tenuiflora are fairly constant. The position of the stamens, 

 the structure of the calyx, the loose inflorescence, the position of the leaves (that is mostly in 

 a basal tuft) are the characters which may be weighed as most dependable. It seems impracticable 

 to delimit proper varieties or subspecies. Instead a number of the extremes have been described 

 in as nearly precise terms as may be, and accorded varietal names for the sake of convenience. 

 Only by some such means can the extent and diversity of the complex contained mthin the species 

 be at all adequately presented in this treatment. Gilia tenuiflora, proper, representing the taxo- 

 nomic type, is here limited to the South Coast Ranges and is segregated geographically, although 

 two varieties, var. arenaria and var. latiflora, occur on the borders of its territory. 



It would appear from a study of Gilia tenuiflora as a whole that reduction or increase in plant 

 size, development of glandulosity, marked leof-segmcntation, variations in distribution of pu- 

 bescence, reduction in corolla size or striking elongation of corolla-tube may have arisen inde- 

 pendently at many different points in the series and thus have brought into close simulation 

 elements of different origin genetically. The forms described varietally below may not be in 

 every case, as to entire content, natural races or only partially such, but a descriptive treatment 

 is needed because of the circumstance that the differences among the several forms are so great 

 in their extreme expression. Such extremes are more unlike than some units of the genus accepted 

 as species. These varieties are, to be sure, of unequal value, as will appear from the manner of 

 treatment. Moreover, all seem to merge by intergrades one with another, mostly on a broad front. 

 They are here described as follows ; 



Var. arenaria (Benth.) Jepson comb. n. Stems 1 or several from the base, ascending, 1 to 5 

 inches high, these and the pedicels and calyx densely clothed with short spreading gland-tipped 

 hairs ; leaves pinnately divided to shallowly toothed, or the leaves merely toothed above or sub- 

 entire, % to 1^2 inches long, mainly basal; flowers racemosely cymose, the pedicels short (Vo to 

 7 lines long) ; corollas 4% to 5 lines long. — Sand dunes, Monterey Bay: Seaside near Del Monte, 

 Heller 6651. The type collection, made by Douglas, is at Kew. It consists of a single dwarf plant 

 two inches high. It is exactly matched as to characters by Elmer 3556, Pacific Grove. The leaves 

 in the type resemble in miniature a double-edged woodman's saw; the basal leaves are glabrous 

 and non-glandular; the stems, pedicels and calyces are clothed with short gland-tipped hairs. 



Var. esllis (Gray) Jepson comb. n. Stems % to 2 feet high, narrowly paniculate or often 

 bushy-paniculate, the brauehlets slender or filiform; leaves small, mainly basal, 1 to 3 times pin- 

 nately parted with the segments mostly long-linear, or the blades rarely entire ; corolla 3 to 5 lines 

 long, its tube very slender, little exceeding the calyx (mostly 1^4 times as long), the limb broad 

 (2 to 4 lines) relatively to the size of the flowers. — Mountain slopes, mesas and valley flats, mostly 

 800 to 5500 feet: cismontane Southern California. 



The prevailing form of the Gilia tenuiflora complex in cismontane Southern California is var. 

 exilis. Its flowers display corollas with the throat gradually but strongly expanded and tlie tube 

 about IVa times as long as the calyx. Var. exilis intergrades freely with var. grinnellii of the San 

 Gabriel Mts., remarkable for its extremely long corolla-tube, and' with forms of var. sinuata of 

 the Mohave Desert, the intergrades as easily referable to one named form as the other. This form, 

 var. exilis, is noteworthy for its altitudinal range; it occurs at Mission San Luis Eey at 100 feet, 

 on Tahquitz Peak in the San Jacinto Mts. at 8000 feet, and on the summit of Mt. Piiios, 8000 feet. 



