THE SALIX LISTS IN THE 'LONDON CATALOGUE.' 465 though degraded by Wimmer and Andersson, and retained as a mere synonym of S. triandra L., and treated in the same way by Dr. White, who regarded it as the equivalent of Andersson's p. concolor 1. latifolia (though it does not agree with Andersson's description), demands fairer consideration. The features which Borrer, who drew Smith's attention to the form, observed are the smaller growth of the shrub (12-15 ft. high), the shorter leaves (1^-2^ in.), which are ovate to ovate-lanceolate from a rounded base, their sides being nowhere parallel," paler beneath, but hardly glaucous. To this description has been added, " young twigs not furrowed"; "stipules larger and more rounded " (Syme, E. B. viii. 215), both good points. Having had the opportunity of watching S. Hoffmannimia Sm. growing side by side with S. triandra type, I am able to confirm all these features, and may say that, whereas S. triandra is a rather lax tree, ascending on the banks of the E. Stour, Dorset, to 25-30 ft., with dark green foliage, 8. Hoff- manniana Sm. on the same river-bank is a small shrubby tree forming a dense head, 10-12 ft. high, with more numerous slenderer twigs, often flattened above the leaf-axils but not grooved, and with foliage of a yellower green colour and a thinner texture. The question, to my mind, is not whether it is a variety of S. triandra, but whether with all these points of difference it does not deserve the specific rank Smith accorded it. Contrast, for instance, the very slight distinctions between S. fragilis L. and var. hritannica F. B. White. I have failed to find in the field the numerous intermediates that Syme refers to ; though it is sometimes difficult to decide on incomplete herbarium specimens. But if, as Dr. E. de Crespigny stated, 8, Hojfmanniana is the only (^ triandra form by the Thames, it is likely that crosses would occur between this and S. triandra 2 » and intermediates might in this way be produced. It is curious that the subspecies (as I am disposed to rank S. Hoffmanniana Sm.) should almost invariably be male. It has doubtless been propagated by cuttings, being frequently of use in strengthening quickset hedges and preserving river-banks. 1396 pentandra X alba [hexandra Ehrh.). In my list the figure 3 placed after this hybrid was intended to include the two counties, Edinburgh and Forfar, mentioned in the Eevision, and a sample from Thornhill, Dumfriesshire, which Mr. J. Fingland informed me Dr. White had named S. hexandra much on the alba side ; and he kindly supplied me with a plant. This plant has developed into a handsome form of S. alba, with no apparent trace of S. pentandra in it. The comital numbers should therefore be 2 ? rather than 3. 1396 pentajidra x fragilis {ciispidata Schultz). This has been discovered at Wybunbury Bog (No. 51, Set of British Willows), Cheshire, by Capt. A. Wolley Dod ; also by the Eev. A. Ley at Pontrilas, in Herefordshire ; the comital number is therefore 3 instead of 1, as in my list. 1397 fragilis Linn. The exact distribution of the type and variety still wants working out. Dr. White knew of three counties only for the type at the date of the Revision ; to these may be added Derbyshire (see No. 2, Set of British Willows).
