368 CONTRIBUTIONS FROM THE NATIONAL HERBARIUM. 



heavy, paniculate by simple racemes distributed along the axis, the lower branches 

 moderately lengthened, excurved-ascending; peduncles 3 to 8 mm. long, naked except 

 at the summit or with one or two lateral subulate bractlets; heads campanulatc, 

 including disk 8 to 9 mm. long; tegules a no t manifestly glutinous, the basal ovate, the 

 others oblong, ovately narrowed to the apex, broadly green-keeled (the two green 

 ridges nearly coalescent); florets 22 to 34, those of the disk 6.5 to 7 mm. long, the 

 achenes with scanty delicate, strigoso hairs; rays about 8, oblong, 3.5 mm. long, of 

 a rather deep yellow. 



The description is based chiefly on an abundance of fresh material from the dunes 

 and flat sands about the head of Lake Michigan: 



Indiana: Dune Park, Porter County, Steele 167, 168, September 17, 1909; Pine, 

 Lake County, Steele 177, September 20, 1909. 



Other material, doubtless the same, is as follows: 

 Indiana: Pine, Lake County, E. J. mil, September 13, 1879 (Herb. Univ. 111.); 



Millers. Lake Countv, L. M. Umbach, September 3, 1897 (Nat. Herb.). 

 Michigan: St. Joseph, Henry Gillman, October 10, 1872 (Gray Herb.); South 

 Haven, L. IT. Bailey, September 24, 1880 (Nat. Herb.) 



All the Gray Herbarium material referred to S. raecmosa gillmani (including the 

 Gillman specimen mentioned) was courteously loaned for comparison. The speci- 

 mens perhaps all belong to the species here described, but no one of them is a normal 

 plant. The Gillman example is broken off above and abnormally branching below, 

 the base not shown. There are besides this a garden specimen "from Mr. Gillman 's 

 plant," specimens grown from Lake Superior roots sent by Boott, and a Kew speci- 

 men supposed to be from the same source; also a sheet from Tobermory, Bruce 

 Peninsula, southwestern Ontario, collected by John Macoun (same in Nat. Herb.). 

 The cultivated plant is greatly altered, the leaves of the sterile tufts being much 

 elongated, and the panicle, in one case especially, much overgrown. The very im- 

 perfect Macoun specimens show also the exaggerated panicle, but lack the basal 

 features. 



Disregarding the Macoun plant, it seems nearly certain that all the material here 

 in view is of the same species, in spite of the startling contrast between the garden 

 specimens and the Indiana material, especially the flat sand form. The garden speci- 

 men from Gillman's plant is so much like the Lake Superior cultures as to leave no 

 ground for a separation. At the same time the original Gillman specimen can, as 

 regards its characters, be perfectly well understood as cospecific with the Indiana 

 material, while its habitat is the same and location adjacent. But evidently the 

 Indiana specimens represent the normal form of the species, and the type conception 

 must therefore be as hero defined. Nontypical features will then be: The elongation 

 of the leaves of the sterile tufts (up to 20 and sometimes almost 30 cm.) and the sharp- 

 ening of their apices, as shown in all the garden specimens, the deep, sharp toothing 

 of the upper leaves, as shown in the largest Lake Superior specimen, the expansion of 

 the panicle into a brush of slender branches, as shown chiefly in the specimen last 

 mentioned, and in a measure the reduction of the thickness of the leaves. One of the 

 Indiana plants, however, has the upper leaves pronouncedly serrate and there is con- 

 siderable variation in the thickness of the leaves, the flat sand plant tending to have 

 them thicker than the dune plant. The peculiarities of the garden plant being merely 

 developmental do not require taxonomic notice. 



Under this new light it is hardly possible to regard the lake sand plant as a sub- 

 species; but in point of relationship S. gillmani is decidedly nearer to S. randii 

 than it is to S. racemosa. From the latter it is separated by its more robust habit, 

 fewer, broader, blunt, more coarsely and strongly reticulate leaves, and its rather 

 larger heads with broader tegules, and by its arenicolous habitat. From S. randii 



«I beg leave to introduce this term in place of the cumbersome "bracts of the 

 involucre." Latin tegula, tegulae, a roof tile, from root teg-, t» cover. 



