130 NATURAL HISTORY OF PLANTS. 



have sometimes been ascribed to Ranuncidacea. But at the 

 same time it has been found impossible to ignore the close 

 affinities between Sanguinaria and Jeffersonia, which only really differs 

 in its unicarpellary gynaeceum. Hence the other Berberidacea, too, 

 border closely on Papaveracea. True they lack milky juice and pos- 

 sess a peculiar colouring matter instead. Their anthers often open 

 by valves, though sometimes by longitudinal clefts. But we have 

 shown that in the latter case the anthers are really introrse, despite 

 contrary appearances, while those of Papaveracece are generally ex- 

 trorse. It is useless to turn to the number of carpels, usually single 

 in Berberidacea but numerous in Papaveracea, since the Lardiza- 

 balece, all pluricarpellary, have been placed among the former, together 

 with PJrythrospermea, whose carpels are united into a one-celled 

 ovary, quite like that of a Poppy. But there is one other difference 

 between Papaveracece and Berberidacece (including Podojjhyllece). The 

 number of stamens, when definite, is a multiple of two in the former ; 

 while in the latter it is a multiple of three, the androceum, consist- 

 ing of deduplicated trimerous verticils. Epimedium, whose species 

 have nearly all dimerous flowers, offers the only exception, and here 

 the valvate anthers are characteristic. 1 By the Fumariece, with the 

 androceum definite, the Papaveracece no doubt approach the 

 hexandrous Cruciferce ; but the stamens are not tetradynamous ; and 

 even when, as in Glaucium, the fruit is a siliqua with a false dissepi- 

 ment, the seeds have a fleshy albumen which is absent in Crucifers. 2 

 The Poppies come especially close to Nymphaacece and Sarracenece. 

 But the latter group have as many petals as sepals ; not as in 

 Papaveracece, a calyx and two corollas, whatever may be the sum 

 total of the perianth leaves ; and the Nymphceece have indefinite petals 

 arranged along a continuous spiral instead of in whorls ; while their 

 seed has a double albumen. We find too among the orders with a 



1 J. G. Agaedh (Theor. Syst., 72, t. 5, figs. structura seminum quoque comparata, typuin 



6, 7) compares Papaveracece chiefly with Nan- omnino diversum mihi indicat." (Ag., loc. cit., 



dinecB, which he separates from Lardizabalece 73.) B. Mirbel (in Ann. Sc. Nat., ser. 1, vi. 



and Berberidacece : " Papaveracece sunt Nan- 266, t. 11) long ago established the chief points 



dinece carpellis pluribus, in pistillum unicum con- of difference between Papaveracece and Cruci- 



junctis. fierce ; and when we study the floral symmetry of 



z " Gemmulse in Papaveraceis ita positae sunt the latter order, we shall see in what respects 



ut raphe lutus superum gemmula? horizontalis aut many botanists have found it comparable with 



erectiusculaj servet, micropyle infera deorsa. Gem- the former, 

 mulas contra in Cruciferis epitropas video, quod, 



