4 
seeds, which fill nearly the whole capsule, it often escapes observa- 
tion, and many of these genera have repeatedly been described as 
having an unilocular fruit. 
From the latter consideration it has been usual, however, to refer 
these supposed unilocular genera to Primulaceee, an order perfectly 
well marked by a very different and a much more important character, 
that of the situation of the stamens, which are opposed to, not alter- 
nate with the petals, showing them to belong to the inner set of a 
double series of stamina, whereas in Scrophularineae the stamina 
are constantly alternate with the petals, and if the normal number be 
really double that of the petals it is the inner series that is con- 
stantly abortive, leaving, as far as has been hitherto observed, no 
trace in any instance, unless, indeed, the deeply cleft disk, described 
by Martins as encircling the ovarium in Herpestis lanigera, or the 
scales in the tube of the corolla of Artanema, be considered as abor- 
tive stamina. 
Schwenckia, hitherto referred to Scrophularineae, has the stamina 
constantly opposed to the petals, as observed by Dr. Lindley upon 
living specimens, and which at his suggestion I have verified upon 
dried specimens of several didynamous as well as pentandrous spe- 
cies,"*^ It is, therefore, unquestionably referable to Primulacece, and 
the glandular processes alternating with the petals may be considered 
probably as abortive stamina of the inner series, analogous to those 
of Samolus. 
It may be scarcely necessary to mention Gentianeje among the 
affinities of Scrophularineae, as their symmetrical flowers and the 
position of the carpels forming the fruit, which when only two in 
number are perpendicular to the axis, not parallel to it, as in Scro- 
phularineae, mark them out at once.f Disandra, an anomalous 
genus of Scrophularineae, has something the habit of Villarsia, but I 
have not had an opportunity of examining sufficiently good speci- 
mens to form any opinion as to its real situation. 
The limits of the order, if it be thus correctly defined and marked 
out from so many neighbours upon whose territory it appears to 
encroach, still remain very wide, and embrace groups considerably 
removed from each other in many points of view, but which yet 
appear too closely allied to be separated otherwise than into tribes or 
suborders. Of these I have, in the above-mentioned paper, pro- 
posed twelve, some of which may be thought natural, others are 
certainly purely artificial, but from the imperfect materials before 
me I have been unable to circumscribe them more satisfactorily. 
A few remarks which I have now to add upon each will show 
some of the difficulties I have met with, and may induce other 
* The fifth stamen is, however, generally sterile, 
t See Lindl. Introd. 2'i9. 
