18 JOURNAL OF THE ARNOLD ARBORETUM [vol. ii 



run: foliis acutiusculis . . . margine snl)revoluti.s. The type of it came 

 from '' Durliam County." To this statement Gray (Sal. Bor.-Am. p. ^'-i) 

 has added '' New Jersey." So far as I know tliere are two counties of this 

 name in North America, one in Ontario, the other in North CaroHna. In 

 New Jersey lliere is only a village named Durham, Andersson's form prob- 

 ably came from Ontario, certainly not from North Carolina. lie says of 

 it: '*Var:eta>, ut etiam nostra, S. gracilis ' S. rosmarinijolia Barratt ' in- 

 scripta, mulio magis ad S, rosviarinifoliam veram accedit. Ohslant autem 

 cai)sulae longissimc pedicellatae." Of this form I have had a fragment 

 before me, and on the photograi)h of the type of S. gracilis there can be seen 

 part of the type of var. rosmarinoidcs. 1 cannot distinguish it from typical 

 gracilis which has entire or more or less serrate leaves. In 1807 Andcrsson 

 dees not even mention var. rosmarinoidcs or the specimen from Durham 

 Comity. He refers to S. gracilis a specimen from '' ad Milwaukee (Laj)- 

 haui)," and this fact proves that he also referred to his gracilis specimens of 



a more soutliern origin. 



In 18G8 Aiidersson somewhat changed his opinion by describing besides 

 var. gracilis a var. angustijolia without mentioning var. rosmarirtoidcs. The 

 type of var. angvstifolia is " S. rosviarinijolia Barratt (lib. Hook.) et Hook. 

 Fl. Boreal. -amer. 2. p. 118 pp." It is characterized by '* foliis anguste 

 lanceolatis fere linearil)us margine tenuiter serruhitis v. subintegris " and 

 'he says of it: "Ilaec quae regiones magis septenlrionales praecipue incolit, 

 nostrae S. rosmarinifoliae sat similis, sed differt foliis subscrralis, capsulis 

 longius pedicellatis et habitu! Ad sequentem transitum evidcnlissimum 

 efTccit et ab ea difhcillime distinguenda; e contrario in formas minus lati- 

 folias et rigidas S, pctiolaris sensim confluit." The "" following '' variety is 

 var. gracilis which, indeed, can l)e distinguished only " difficillime " or 

 better not at all from var. angustijolia. This last form certainly is identi- 

 cal with var. ro^marijioidcs of 1858, and this is another instance of Anders- 

 son's changing of names without even ([Uoting again his former denomina- 

 tions. Now, if we take gracilis for a variety of 8. pctiolaris we must use 

 the name ro.srnarinoides for it. Barratt, to be sure, had already made a S. 

 pctiolaris var, jS angvstata in 1840 (Sal Am. sub no. 9), and he says of the 

 typical pdiolaris "there are several varieties of it; some with narrower 

 leaves," but he had in mind only forms of Pennsylvaiua and New York. 

 Therefore, this var. angustata refers to narrow leaved forms of the type 

 whicli have longer fruiting aments with more crowded fruits with pedic(*Is 

 up to 3 mm. in length. Such forms have been distributed l)y Bcbb in 

 his Herb. Salic, under nos. 28 and 29; they seem to unite the ty[>e with var. 

 rotmarinoidcs. Of this variety the type shows a fewstomata in the upper 

 epidermis of the leaves, something Avhich I have not o])servevI in other 



specimens. 



In the forms of typical var. rosmarinoides the under surface of the le; 

 is always glabrous or almost so (with exception perhai)s of the very youngest 

 leaves) while the upper surface is more or less hairy. TIhtc occurs, how- 

 ever, in Maine and Quebec a form of which the young leaves (at least partly) 



