For the restoration of sensitiveness, and to meet working conditions, the lower end of the cut stem is mounted on a T-tube, with funnel-attachment and exit-tube, as shown in Fig. 30. The other two cut ends--of the stem and of the petiole--may be covered with moist cloth or may be closed with collodion flexile to prevent rapid evaporation and drying up of the specimen. A slight hydrostatic pressure maintains the specimen in a moderately turgid condition. A preparation thus made is insensitive at the beginning, but if left undisturbed it slowly recovers its excitability. The history of the depression of excitability after shock of preparation and its gradual restoration is graphically illustrated by a series of records made by the plant (Fig. 31). The petiole-pulvinus preparation thus made offers all facilities for experiment. Owing to its small size it can be easily manipulated; it can be enclosed in a small chamber and subjected to varying conditions of temperature and to the action of different vapours and gases. Drugs are easily absorbed at the cut end, and poison and its antidote can be successively applied through the funnel without any disturbance of the continuity of record. In fact, many experiments which would be impossible with the entire plant are quite practicable with the isolated leaf. The arrangement for taking records of response is seen in Fig. 30, which is reproduced from a photograph of the actual apparatus. For recording the response and recovery of the leaf under stimulation, I use my Resonant Recorder fully described in the ‘Philosophical Transactions’ (1913). The petiole is attached to one arm of the horizontal lever. The writer, made of fine steel wire with a bent tip, is at right angles to the lever, and is maintained by electromagnetic means in a state of to-and-fro vibration, say, ten times in a second. The record, consisting of a series of dots, is free from errors arising from friction of continuous contact of the writer with the recording surface. The successive dots in the record at definite intervals of a tenth of a second also give the time-relations of the response curve.