NORMAL KNEE-JERK. 17
there was no appreciable friction in the apparatus, there could be no doubt that it gave a blow of definite force. The blows were delivered at intervals of fifteen seconds ; therefore, the variations could not be due to a wearying of the muscle. Moreover, the knee-jerk was often greater at the end of a series of experiments than at the beginning. The only chance for error seemed to lie in the possibility that the position of the leg was changed slightly from time to time, and that the hammer did not strike the ligament at ex- actly the same place each time. This question was carefully studied and we were unable to find that there was any such change of position. Moreover, we discovered that it made no appreciable difference in the extent of the knee-jerk whether the hammer struck exactly the middle of the ligament, as we always tried to have it do, or a little above or below that point. Having ruled out all possible sources of error, we were compelled to conclude that the varia- tions which we saw were due to changes which oc- curred within the individual and which reenforced the action of the mechanisms which produce the knee- jerk. Succeeding experiments proved that there was no lack of reenforcing influences.
The variations seen were compared with strongly reinforced knee-jerks. — Having once assumed that the variations which we had seen were due to some reenforcing influence, we had the curiosity to compare the largest of the knee-jerks, obtained when the subject was entirely quiet, with those which should result from some of the vigorous forms of reenforcement, described by Mitchell and Lewis, such as clinching the hands or clinching the teeth. The results of a few experiments, in
