28 JOUHNAL OF THE ARNOLD ARBORETUM [vol. i 



parte. — Hough, Handb. Trees 76, figs. 89-90 (1907), pro parte. — Britton 

 & Shafer, N. Am. Trees, 186 (1908), pro parte. — Ball in Elys. Mar. iii. 

 21 (1910), pro parte. — This variety has been well treated by Glatfelter 

 (1897) who, however, did not separate it from the t>^)ical ^S. lungipcs. The 

 following statement of this excellent observer has certainly not lost any- 

 thing of its value since he wrote it 21 years ago: " Whether or not S. longi- 

 pcs should be divided into several forms or varieties is a question, owing to 

 insufficient accumulation of material, not yet determinable. In view of 

 what we know at the present day of the instability of some of our species of 

 Willows, it does not seem to fill the present denuuids of the scientific idea, 

 to set up a numl)er of forms from a limited num1)er of herbarium si>ecimens, 

 and, besides, incomplete as we often find Ihem. Without careful observa- 

 tions, such work nmst inevitably i)rove defective. The plant must be seen 

 its habitat and in quantity before a just conception of all its characters 



m 1 



can be formed. In this way the supposedly good form will often vanish, to 

 be supplanted by easily recognizable variations." Glatfelter did not have 

 very copious material of the typical S. lorujipes at his disposal, neither was 

 he well ac(|uainted with the forms I refer to var. venulnsa. So far as I can 

 see, these varieties inhabit diiTerent geographical areas wliirh, however, 

 meet together in certain regions, and it is mostly in these that we may 

 observe intermediate forms. The range of var. Wardii is not yet sufHciently 

 kno^\^l. The type was found by Ward near Washington, D.C., ^diere it 

 ap])arcntly reaches the most northern point of its distribution. The north- 

 ern borderline thenccfrom runs along the Potomac River to zVHegany 

 County, Maryland. The next station from which I have seen a specimen 

 is Fpsimr County in W. Virginia, and it has been reported from Lexington 

 in Fayette County, Kentucky. I have not yet seen specimens or reports 

 of it from AVest Virginia, Ohio, or Indiana. In Illinois it was collected 

 in St. Clair and Madison Counties, and it seems to show its best develop- 

 ment in :Missouri, from which I have seen specimens from the following 

 counties: (east to west) Marion, Ralls, Pike, St. Louis, St. Francois, Madi- 

 son, Iron, Washington, Phelps, Shannon, Carter, Howell, Wright, Chris- 

 tian, Taney, Stone, McDonald, Newton, Jasper, Jackson. In Kansas I 

 have seen it only from the southeast corner (Cherokee County), a speci- 

 men from the Kiowa County being doubtful. From Oklahoma I have had 

 specimens before me from Ottawa, Drage, Rogers, Creek and La Flore 

 Counties in the northeast and east, while those from Comanche County 

 in the southwest may probably better be referred to var. venulosa. The 

 southern borderline of its range apparently runs from La Flore County 

 in Oklahonui through adjacent western Arkansas (Howard County) and 

 northwestern Arkansas (Benton and Carroll Counties) and southern Mis- 

 souri, and northern Tennessee (Davidson and Robertson Counties); it has 

 been collected by Harbison as far south as northeastern Mississippi (on the 

 Tennessee River, near luka, Tishomingo County), and ISTohr cites it from 

 the northwestern corner of Alabama (Lauderdale County, Teimessee River). 

 Farther to the cast I have seen var. Wardii from Natural Bridge, James 



