2

Scientific Memoirs by

filmenta duo suprema sterilia into "dix étamines libres, huit fertiles, les deux
supérieures stériles." This important change, though afterwards adopted by
DeCandolle, in connexion with whose reference it will be discussed, was not
altogether justified, for Poiret does not say that he had been able to examine a
flower.

      The next reference to Afzelia is in Rees' Cyclopœdia i. (1819). Smith,
we know, wrote botanical notices for this work. But he did not write the article
Afzelia. The Latin of the author of that article had become rusty; the expres-
sion Summo maximo, applied to the petals, is rendered "with a very large head ";
the staminal character is translated " the filaments are two, superior, sterile,"—a
repetition of the misconception embodied, even if not intended, in Persoon's
notice. The article Pancovia in Rees' Cyclopœdia xxvi. was, however, so De-
Candolle tells us, written by Smith. It is there suggested that Pancovia and
Afzelia may be congeneric. This suggestion DeCandolle tentatively adopted.
A year later Sprengel, Syst. Veget. cur. post. 170 (1827) did so definitely. He
was not, however, right; the species now proves to be an Erioglossum
(Sapindaceae ). Sprengel's reference to A. africana, in Syst. ii. 345 (1825), is
negligibly brief.

      The account of Afzelia in DeCandolle's Prodromus ii. 507 (1826) calls for
a closer scrutiny than it usually receives. It describes the seed more fully than
Smith had done; DeCandolle states that he had seen a seed. It differs further
in adopting Poiret's amplification of Smith's character filamenta duo suprema,
sterilia, into stamina 10quorum 8 fertilia 2 super. sterilia; Persoon and
Rees, by their misinterpretation of Smith's original phrase, had rendered some ac-
tion of this kind necessary. As DeCandolle did not see a flower, this amplification
is only a deduction from Smith's imperfect characterisation. Though at first sight
a natural enough deduction, it is not the only possible one; in this case it is not even
the most probable. It does not follow, because his plant was referred to Decand-
ria, that all ten stamens were represented in Smith's flower. The fact that the
uppermost stamens were represented by a priori of staminodes instead of by one
or some other odd number should, on a priori grounds, indicate that the
vexillary stamen was unrepresented even by a staminode. Then Smith's own
suggestion that Pancovia, which Willdenow had referred to Heptandria, might be
congeneric with his Afzelia, should imply that in Afzelia also there were seven
perfect stamens. The unimportant mistake which began in Persoon's Synopsis,
of citing the name Afzelia africana as having been published in Smith's original
account of the genus, recurs in the Prodromus. The error is not without its use;
it has been freely copied and betrays those writers who have cited Smith as their