It is, however, difficult to obtain an absolutely cylindrical specimen, as it always tapers in one direction. The conductivity towards the tip of the root is not exactly the same as that in the ascending direction. It is therefore difficult to fix an absolutely neutral point, but a point may be found which approaches this very nearly, and on stimulating the stalk near this, a very interesting diphasic variation has been observed. In a specimen of cauliflower-stalk, (1) stimulus was applied very much nearer A than B (the feeble disturbance reaching B was negligible). The resulting response was upward and the recovery took place in about sixty seconds. [Illustration: FIG. 26.--DIPHASIC VARIATION] (2) Stimulus was next applied near B. The resulting response was now downward (fig. 26, _b_). (3) The stimulus was now applied near the approximately neutral point N. In this case, owing to a slight difference in the rates of propagation in the two directions, a very interesting diphasic variation was produced (fig. 26, _c_). From the record it will be seen that the disturbance arrived earlier at A than at B. This produced an upward response. But during the subsidence of the disturbance at A, the wave reached B. The effect of this was to produce a current in the opposite direction. This apparently hastened the recovery of A (from 60 seconds to 12 seconds). The excitation of A now disappeared, and the second phase of response, that due to excitation of B, was fully displayed. #Positive after-effect.#--If we regard the response due to excitation of A as negative, the later effect on B would appear as a subsequent positive variation. In the response of nerve, for example, where contacts are made at two surfaces, injured and uninjured, there is sometimes observed, first a negative variation, and then a positive after-effect. This may sometimes at least be due to the proximal uninjured contact first giving the usual negative variation, and the more distant contact of injury giving rise, later, to the opposite, that is to say, apparently positive, response. There is always a chance of an after-effect due to this cause, unless (1) the injured end be completely killed and rendered quite irresponsive, or (2) there be an effective block between A and B, so that the disturbance due to stimulus can only act on one, and not on the other. I have found cases where, even when there was a perfect block, a positive after-effect occurred. It would thus appear that if molecular distortion from stimulus give rise to a negative variation, then during the process of molecular recovery there may be over-shooting of the equilibrium position, which may be exhibited as a positive variation.