MAGNOLIACEAE. — MAGNOLIA 401 



few words must be said about the source of this confusion. The first mention of 

 Asiatic Magnolias is found in Kaemjifcr's Amoenitates cxoticae, where he describes 

 three species under the names (1) Sini . . . vulgo Kobus, which is M. kobus 

 De Candolle, (2) Mokkwuren frutex tulipifer and (3) Mokkwuren flore albo. 

 In 1791 Banks pubUshed the excellent and very characteristic drawings by 

 Kaempfer of these three species, but made the unfortunate mistake of inter- 

 changing the plates of the last two species, referring plate 43 (Mokkwuren 1.) 

 which represents Mokkwuren flore albo to Mokkwuren [frutex tulii)ifcr], and 

 plate 44 (Mokkwuren 2.) which represents Mokkwuren frutex tulipifer to 

 Mokkwuren flore albo. In the same year Desrousseaux in Lamarck, Ency- 

 clopedie methodiquc drew up descriptions of these two species under the names 

 M. denudata and M. liliflora; his descriptions are bavsed entirely on Kaempfer's 

 plates, except as to the color of the flowers which he took from Kaempfer's descrip- 

 tion, accepting the quotations as given by Banks; this caused him to attribute 

 to M. denudata red flowers and to M. liliflora white flowers. He apparently 

 did not compare carefully enough the plates with Kaempfer's original de- 

 scription; if he had he would have detected Banks' error. Kaempfer describes 

 Mokkwuren frutex tulipifer . . . flore Lilio-narcissi rubente as similar to Sini 

 . . . vulgo Kobus, the flowers of which he calls " Tulipam Liliumve album 

 vulgare petalorum numero et magnitudine exprimentibus . . . " ; these words 

 undoubtedly refer to a flower with six petals, as plate 44 shows, while in de- 

 scribing Mokkwuren flore albo he says " novemque plerumque petalis," exactly 

 as shown in plate 43. This proves conclusively that Kaempfer did not attribute 

 red flowers to his drawing published by Banks as plate 43, as one is lead to believe 

 from Bank's quotation, and that this plate represents Mokkwuren flore albo, and 

 therefore has white flowers. Maximowicz (1. c.) apparently had arrived at the 

 same conclusions, as he quotes under M. conspicua: " Mokkwuren Ic. Kaempf. 

 t. 43 — Moldcwuren fl. albo novem plerumque petalis cet. Kaempf. Amoen. 845," 

 but he does not mention Desrousseaux's names. 



The fact that Desrousseaux describes the flowers as red instead of white, owing 

 to a wrong citation in the synonymy, is not a sufficient reason to reject his name. 

 De Candolle made the same mistake in describing M. kobus; he quotes M. gracilis 

 Salisbury as a synonym and describes the flowers as red, taking the description of 

 the color from the colored plate of M. gracilis. There are also numerous other in- 

 stances where the color of the flowers has been incorrectly given in the original 

 description without afTecting the validity of the name. The acceptance of the 

 name M. denudata fortunately makes it unnecessary to decide whether M. prccia 

 Correa or M. conspicua Salisbury is the older name. 



Magnolia denudata, var. purpurascens Rehder & Wilson, n. comb. 



Magnolia conspicua, var. purpurascens Maximowicz in Bull. Acad. Sci. St. 



Petersbourg, XVII. 419 (1872); in Mel. Biol. VIII. 509 (1872). 

 Magnolia obovata Keisuke Ito, Fig. Descr. PI. Koishikawa Bot. Gard. I. t. 8 



" Sarasa-renge " (non Thunberg) (1884). 



Western Hupeh: Changyang Hsien, woods, thickets and open 

 country, alt. 1300-1800 m., common, April 11 and September 1907 

 (No. 278; tree 5-18 m. tall, 0.3-2 m. girth, flowers rosy pink, fragrant) ; 

 Hsing-shan Hsien, woodlands, alt. 13-1500 m., common, September 

 1907 (No. 373; tree 5-18 m. tall, 0.3-2 m. girth); Fang Hsien, moist 



