162 
H. F. Wernham. 
similar to that by which Globulariaceae descended from Selagineae 
{supra). Evidence of arrested development of loculi is not 
wanting among Verbenaceae, and the nucules in the fruit of some 
Labiatae may number three, two, or one only, owing to abortion. 
Apart from the characters of the gynaecium, Phryma is 
undoubtedly verbenaceous. 
Myoporaceae includes less than 100 species, chiefly Australian 
and Polynesian. Their usually anisostemonous andrcecium with 
didynamous stamens and more or less irregular flowers entitle them 
to a place among the higher Tubiflorae; but their exact position 
within the latter group has been regarded as questionable. Eichler 
and Bentham and Hooker associate them with Diovulatae; Engler 
assigns them the rank of a separate sub-cohort of Tubifloras, placing 
them after his multiovulate groups and widely removed from sub¬ 
cohort B (Diovulatae). The question turns principally upon the 
number and arrangement of the carpels, seed-chambers, and ovules. 
It is probable that in all cases the ovary is primarily bicarpellary ; 
and in all the genera except Myoporum —i.e., in nearly three-fourths 
of the total number of species—the ovary is constantly bilocular. 
In Myoporum (20 species) the number of loculi is indefinite (2 
to 10), and this is due to secondary septation; the motive of the 
latter is reflected in the fact that only one ovule is present in each 
chamber of the multilocular forms, while in the bilocular forms 
more than one may be found. In Eremophila, the largest genus 
(40 species), there may be four, six or eight ovules in each of 
the two loculi, superposed in pairs; there are rarely so few as two. 
Pholidium, with 15 species, is entirely typical of Diovulatae in 
having two ovules (rarely one) associated with each carpel. 
The occurrence of indefiniteness in ovule-number may be 
regarded as favourable to the inclusion of Myoporaceae in Multi- 
ovulatae; but the clear tendency to schizocarpy by secondary 
segmentation is still more favourable to their classification with 
Diovulatae. The position of Myoporaceae relatively to Diovulatae 
will then be seen to be comparable with that of Nolanaceae relatively 
to Boraginaceae, in the Transitional Group (see chapter V); for in 
either case the ovule-number is indefinite in the questionable group, 
but the latter is linked with the family having definite ovule- 
number by the character of schizocarpy. 
Other considerations support the suggested affinity of Myo¬ 
poraceae with Diovulatae. They recall the verbenaceous Avicennia 
closely in general habit, and the drupaceous fruit is practically 
