the appendages of the corolla appear as tufts of hairs inserted at the base of the torus, remaining after the petals have been taken away. It should be here noted that the whorl which the author described as a calyx may have been in fact an outer series of petals, conforming, as in certain allied species, to the true sepals, which in this case, however, must have been highly caducous.
In 1834 Wight and Arnofct (Prodromus p. 10) under "Miliusa indica (Lesch.)" reproduced A. Le Candolle's leading characters, but assigned them to " Wall. L.n. 5433, " quoting as a synonym Uvaria ciliata, Heyne MSS." There is a specimen at Kew ticketed Uvaria ciliata, Heyne Tenmalej near Courtallam, July 1815," and on the same sheet is a second ticket from which it seems probable that the citation of " Wall. L.n. 5433 " may refer to this specimen or to a duplicate.
In April 1835, Wight got a plant at Courtallam which he ultimately referred to Miliusa indica, and in the following July and August he collected further specimens which together with the April gathering have since (1866-67) been distributed from Kew as No. 34 of the Indian Peninsular Collections. These later gatherings (in part) resemble "Uvaria ciliata, Heyne" so far as can be judged, but Heyne's specimen is not in fruit : No. 34 of Wight has drupes and matches an example from his own Herbarium, without date or original locality, with which is now attached a description in his own band-writing. This description, however, seems to have belonged to thspecimen distributed under No. 33 as "Miliusa montana, Gardner" (which in PL Brit. Ind. i. 86 is reduced to M. indica) and to have been pinned to the sheet where it is now found by accident. A pencil note on the sheet from Wight's own Herbarium (in Sir J. D. Hooker's handwriting apparently) points out that the plant there represented differs from No. 33 by the ripe carpels being sessile and pubescent-tomentose. This form (No. 34 in part) seems to have received from Wight the manuscript name of "M. affinis". Nothing on these sheets can be referred to Alph. De Candolle's M. indica unless we assume—
(1) that the sepals, which in Nos. 33 and 34 Herb. Pen. Ind. are manifest, may sometimes fall off at a very early stage, or
(2) that the sexes are diclinous and that the structure of the androecium differs in the male and female flowers very widely, or
(3) that the arrangements of the floral whorls is remarkably unstable.
How far any parallel to the degree of instability that must be supposed in this case has been observed in other Anonaceae is a question which may be deferred for the present. In the "Illustrations" i, p. 68 (1840), Wight met this difficulty by suggesting that Da,
