slightly warm them; so that if they were injured by the frost in a greater degree than the free vertical leaves, the evidence would be so much the stronger that the horizontal position was injurious.  But we found that when there was any slight difference in the result, which could be detected only occasionally, the leaves which had been fastened closely down suffered rather more than those fastened with very long and [page 287] thin pins, so as to stand from ½ to 3/4 inch above the cork.  This difference in the result, which is in itself curious as showing what a very slight difference in the conditions influences the amount of injury inflicted, may be attributed, as we believe, to the surrounding warmer air not circulating freely beneath the closely pinned leaves and thus slightly warming them.  This conclusion is supported by some analogous facts hereafter to be given.

We will now describe in detail the experiments which were tried.  These were troublesome from our not being able to predict how much cold the leaves of the several species could endure.  Many plants had every leaf killed, both those which were secured in a horizontal position and those which were allowed to sleep—­that is, to rise up or sink down vertically.  Others again had not a single leaf in the least injured, and these had to be re-exposed either for a longer time or to a lower temperature.

[Oxalis acetosella.—­A very large pot, thickly covered with between 300 and 400 leaves, had been kept all winter in the greenhouse.  Seven leaves were pinned horizontally open, and were exposed on March 16th for 2 h. to a clear sky, the temperature on the surrounding grass being -4o C. (24o to 25o F.).  Next morning all seven leaves were found quite killed, so were many of the free ones which had previously gone to sleep, and about 100 of them, either dead or browned and injured were picked off.  Some leaves showed that they had been slightly injured by not expanding during the whole of the next day, though they afterwards recovered.  As all the leaves which were pinned open were killed, and only about a third or fourth of the others were either killed or injured, we had some little evidence that those which were prevented from assuming their vertically dependent position suffered most.

The following night (17th) was clear and almost equally cold (-3o to -4o C. on the grass), and the pot was again exposed, but this time for only 30 m.  Eight leaves had been pinned out, [page 288] and in the morning two of them were dead, whilst not a single other leaf on the many plants was even injured.

On the 23rd the pot was exposed for 1 h. 30 m., the temperature on the grass being only -2o C., and not one leaf was injured:  the pinned open leaves, however, all stood from ½ to 3/4 of an inch above the cork.