296 MR. J. C. WILLIS’S CONTRIBUTIONS TO 
about 2 mm. long, and 14 mm. wide, blue at the further end 
(fig. 22). The position of the essential organs is shown in fig. 23, 
the style being curled round the anthers in the lower lip, where it 
is fertilized. Nearly every flower set a full complement of fertile 
seed. During the early part of June 1893 all the flowers were 
of this type. Towards the end of the month appeared flowers 
with larger corollas, as much as 5 or 6 mm. long; these were 
slightly open at the mouth, but the style still ran into the lower 
lip and was fertilized there. Soon after, flowers like fig. 24 
were found, with the style projecting at the mouth, but it was 
found to have been pollinated in the bud. Finally a few flowers 
were observed like that shown in fig. 25, with the corolla fairly 
large, its mouth wide open, and the lever mechanism in good 
working order. These flowers were fully capable of insect- 
fertilization, and did not pollinate themselves in the bud. No 
insects were seen to visit them, however, which is not surprising 
when we consider their inconspicuousness. 
Later on in the season (August onwards) the plant bore the 
cleistogamic flowers again, without any intermixture of open 
ones. On the whole, as the season advanced, the corollas became 
steadily smaller, till in November they were once more only 
about 2 mm. long ; the flowers, however, were still fully fertile. 
One flower was noticed with the stamens completely aborted. 
It would thus appear that the forms of flower on Salvia Ver- 
benaca vary considerably with the season of the year at which 
they are examined, and this fact must therefore be taken into 
account in considering the forms and varieties of the species. 
From a natural history point of view, the facts described above 
are of some interest as throwing light upon the origin of cleisto- 
gamy. The plants bore flowers with very reduced corollas at 
the beginning and end of their flowering season, and the larger 
forms chiefly in June and July. It would thus seem probable 
that the cause of the variation in size of the corolla is either 
light or temperature, or both together, but the observations made 
gave no evidence as to which of the two was the important 
factor. In many plants with cleistogamic flowers, e. g. Lamiwm 
amplexicaule, these flowers appear at a season when the light and 
temperature are lower than at the time of appearance of the 
normal flowers. In others, e.g. Viola, the cleistogamic flowers 
appear in the brighter and warmer period, but are found usually 
to be on the shaded parts of the plant. The first-named facts 
