MYRTALES. 



828 



size in Plate XCVI, fisjuro 1 , lying across a leaf 

 of Cassia ( mdniinnta Berrv, tlio other markings 

 on the loaf being those of a well-marked leaf- 

 spot fungus. Before descrihing the flower in 

 (h>lail it should be pointed out tliat figures 1, 2, 

 and 4 of Plate X(A'I are from photographs that 

 have not been retouelied in any ])articular. 



As I int(M-pret it, the flower shows tlie foUow- 

 uig characters: Peduncle stout, ciu'vi'd, about 

 4 millimeters long. Calyx ratlier deeply four 

 orfive lobed, tlie lob(>s ovate in outline and llie 

 tips bluntly pointed. Corolla polypetalous 

 composed of foiu' or five petals alternating willi 

 the calyx lobes, long and narrow, seemingly 

 pointed, about twice tlie lengtli of the calyx 

 lobes. Ovary inferior, style long and slen(k'r, 

 probablj' bearing a single tenninal stigma. 

 Stamens 12 in number, exserted, the filaments 

 long and slender. .Vntliers elongate-elliptical 

 in form, two celled, dehiscuig by longitutlinal 

 slits. The stamens may vary in length or tlieir 

 apparent variation may be simpty a feature of 

 preservation. I am not sure that the ap- 

 pearances that I have interpreted as petals are 

 correctly identified, but it is hard to imagine 

 what else tliey can possibly represent. Tlie 

 single slender style is also a feature that may 

 be simulated by a filament. As shown in tlie 

 accompanying restoration, the flower is poly- 

 petalous, regular, and perfect. It is repre- 

 sented as having a four-lobed calyx and four 

 petals, although only three cal3rx lobes and two 

 petals are distinctly seen in the specimen. If 

 four is the correct number then the stamens are 

 three times as numerous as the petals. The 

 reason for the belief that these flowers were 

 capitate or grew in crowded spikes is their 

 small size, narrow petals, and exserted style 

 and stamens — all characters shared by the 

 flowers of the Mimosacea> and Combretacese, 

 the two families whose flowers are most like the 

 fossil. In the Mimosaceje the filameiits are 

 as a rule more sleinhn- and more elongated, as 

 well as more or less united, but in tlie fossil they 

 are free. Tiie anthers are also mucli smaller 

 and less elongated in tlie Mimosacese. The 

 flowers of most of the Conibretacea3 are very 

 similar to tln^ fossil, although the stamens are 

 generally reduced in number to twice the num- 

 b(>r of the petals or of the calj'x lob(>s in the 

 apetidous forms. However, some of tluMnodern 

 forms have three times as many stamens as 

 petals or calyx lobes. The most similar modern 



flowers in appearance that I have been able to 

 find are those of Comhrrtum {/uidruiense Kusby 

 from Bolivia, and in this species the stamens are 

 only eight in number and more exs(>rted. I 

 have submitted the specimen to botanists who 

 are familiar with the flora of tropical Am(>rica 

 and have also compared it with a vast amount 

 of r(H;ent material, and 1 am satisfied that it 

 represents an Eocene mendier of the Com- 

 bretacea>, a family that was apparently well 

 represent(ul in the early Eocene, since I have 

 described from contemporaneous deposits the 

 leaves of three species of Combretum, three 

 species of TenninaLia, one species of Cono- 

 carpus, and both leaves and fruit of a species of 

 Lagunculari.i.. These all serve in a measure to 

 substantiate on(^ another, and a certain amount 

 of confirmatory (evidence is furnished by the 

 petrified wood described by Felix from the 

 European Eocene as Combretacmiimi' and 

 comparetl with the woods of modern forms of 

 Terminalia, Bucida, and the like. Leaves of 

 Terminalia and Combretum have also been 

 described by several authore from the European 

 Tertiary. 



Occurrence . — Lagrange formation (in beds of 

 Wilcox age), IJ miles west of Grand Junction, 

 in Fayette Comity, Tenn. (collected by L. C. 

 Glenn). 



Collection. — U. S. National Museum. 



Genus TERMINALIA Linne. 



Terminalia lesleyana (Lesquereux) Berry. 



Plate I.XXXIX, figure 4. 



Terminalia radobojensis Heer. Lesquereux, U. S. (ieol. 



and Geog. Survey Terr. .\nn. Kept, for 1871, 



suppl., p. 15, 1872. 



Magnolia lesleyana. Lesquereux, Am. Pliilos. Soo. Trans., 



voL 13, p. 421, pi. 21, figs. 1, 2, 18C9, 



Lesquoreiix, The Tertiary flora, ]). 248, pi. 44, figs. 



1-3, 1878. 

 Lesquereux, U. S. Geol. and Geog. Survey TeiT. Ann. 

 Rept. for 1871, suppl., p. 14, 1872. 



Description. — Lesquereux's description, pub- 

 lished in 1869, is as fo flows: 



M. f( liis obovatLs, breviler obtuse acuminatis. in i)otio- 

 linn longe attenuatis; nervo medio valido, transvoraim 

 eroso striate; nervis sooundariis distantibus, irregularibus, 

 secundum margiueni valde curvatia. 



The type material came from Hurleys, in 

 Benton County (formerly the western i)art of 



1 Felix, Johannes, Deutsche geol. Gcsell. Zeitschr., vol. 46, p. 90, pi. 10, 

 flgs. la-c, 1894. 



