38 



MEMOIRS OF THE NEW YORK BOTANICAL GARDEN 



[VOL. 10 



Tabasco: lakeside at Ojo de Agua near Balancan, Mai 1939, Matuda 3136 (F, GH, MICH 

 (type collection of Spathiphyllum lacustre Lund.). Yucatan: cenote at San Jose, Chichen Itza, 

 Jun 1932, Steere 1592 (MICH). 



CULT.: In large pot, Hotel Francia, Oaxaca de Juarez, Mexico, Mar 1940, Bailey 552 

 (BH); Hort. M. Devansaye, Jan 1879 (K) ; Engler 17 (B, C, G-BOIS, G-DEL, GH, L, P, 

 IS); Fragment and outline of type specimen of 8. lan-crolat urn C. Koch, Brown 8,n. (K) ; etc. 



Vernacular names: Veracruz: "chil de gatos"; Oaxaca: "hoja de piedra." 



This is the northernmost species of the genus, extending nearly to 20° N. 

 latitude in Veracruz, Mexico. Some of the material referred here is more or less 

 anomalous, but the species is poorly collected and imperfectly understood. 



S. cochlearispathum was initially described in Hydnostachyon. That genus 

 was differentiated from Spathiphyllum by a unilocular ovary. The ovary, how- 

 ever, is clearly trilocular in all specimens originally assigned to Hydnostachyon, 

 and Schott correctly referred them to his earlier-described genus Spathiphyllum 

 in 1853. 



Engler (Monogr. Phan. 2: 221. 1879) placed S. longirostre as a variety of 

 S. cochlearispathum, which treatment he retained in lf)()8 (Pflansenreich 4 2::n : 

 122). I cannot justify the recognition of var. longirostre. "The type specimen of 

 S. cochlearispathum and of S. longirostre are so similar that they are here treated 

 as conspecific. 



Matuda (Anal. Inst, Biol. Mexico 25: 124. V.)7A) accepted S. lacustn Lund., 

 based upon a single collection (Matuda 3136) from Tabasco. To maintain it 

 apart from S. cochlearispathum seems unwarranted, since the type specimen of 

 S. cochlearispathum and of S. longirostre possess the characteristics stated by 

 Matuda for separating them from S. lacustre. All have five 1 to three ovules per 

 locule and the geniculum varies from three to five centimeters in length. 



A Karwinski specimen (M) from Mexico (locality unknown) is annotated 

 "Spathiphyllum longirostre Schott, ipse." It is very different from the holotype 

 of S. longirostre. If conspecific, these elements must represent the extremes of 

 variation in the species. Though cited here, the Karwinski collection was not 

 considered in preparation of the description of S. cochlearispathum. It resembles 

 most nearly Matuda 3136 from Tabasco, and is similar to two Liebmann speci- 

 mens [s.n,, designated by me M41C and MUD (C)} from Mirador, Veracruz. 



The most interesting, and perhaps most nearly distinct form referred here 

 is a collection from a "cenote" at Chichen Ttza, Yucatan (Steere 1:~>92), with a 

 proportionally wider leaf-blade, a longer petiole and shorter peduncle, and a 

 spathe more shortly decurrent on the peduncle than is typical. When material is 

 available from other "cenotes" in that region, a clearly defined taxon may become 

 apparent and deserve recognition. 



It appears that S. lanceolatum C. Koch is synonymous with S. cochleari- 

 spathum rather than with S. fricdriehstliolii as treated by Engler. This may never 

 be satisfactorily resolved because the holotype of S. lanceolatum was probably 

 burned at Berlin. However, a Kew sheet prepared by N. E. Brown (12 Jul 1878) 

 consists of a drawing (an "outline of the type specimen! in Herb. C. Koch!") 

 and a fragment ("a few ovaries from C. Koch's type!"). The "ovaries" con- 

 tain fifteen or sixteen ovules, and* the dimensions of the outline of the leaf-blade 

 and spathe agree with those of this species rather than of S. frieelrichsthalii. 



Material of Engler 173, distributed under the incorrect identification of 

 S. lanceolatum C. Koch, does not belong here (see under 8. Jcochii) . 



26. S. matudae Bunting, sp. nov. Figure 8. 



Foliorum lamina obliqua, e lanceolata ad elliptico-lanceolatam vel aliquando 



