Of all the human reflexes that may be elicited by appropriate stimuli, that of the patellar lends itself most conveniently to measurement. No doubt the method of recording the latency, amplitude, and refractory period of this reflex from the thickening of the quadriceps muscle is, all things considered, the most satisfactory technique.2 The term "reflex arc" is associated with such ideas as simplicity and invariability, in short, constancy of response. Physiologists, however, who have worked with human reflexes, and specifically with the patellar reflex, have soon discovered that this is not a constant in its time relations or in its amplitude. Our data for the 63 young men of the normal series of 1917, with whom the patellar reflex was measured with identical apparatus and procedure as in the low-diet research, may be examined in connection with this matter of the variability of this reflex. In the first place, it is noteworthy that of the 63 men there were 15 from whom the reflex could not be obtained in measurable amplitude. Every effort was made to adjust the apparatus to exactly the proper height on the tendon, to see that the leg of the subject was in a comfortable position, and that the subject was relaxed, as indicated by the condition of his leg muscles. Our failure was, therefore, not because of lack of time or care in trying to secure the reflex. It was because under our conditions for stimulation and on that date the reflex could not be produced. In several of these cases, when trial was made with the legs crossed and by the usual clinical method, there was some reflex, but of course this was not measurable as to latency or amplitude under such conditions.

The 48 men from whom measurable series of reflexes were obtained gave an average latency (average of all individual averages) of 32σ3 The standard deviation for this series of 48 patellar reflex latencies was 6.6 σ, the coefficient of variability, therefore, being 21 per cent. We may anticipate our results slightly here and point out that this coefficient of variability is larger than a similar coefficient for neuromuscular processes, such as reactions and muscle coordinations, which are usually considered much more complex than the patellar reflex. The average amplitude of the primary reflex for the 48 individuals was 15 mm. This is a magnification of 6 times the extent of actual muscle-thickening. With different subjects the average amplitude varied from 2 to 45 mm. The standard deviation for the series was 11.4 mm., thus showing a coefficient of variability for the series of 76 per cent.1 The time interval separating the two stimuli to the patellar tendon was gradually changed to determine that time separation when the second stimulus failed to produce a measurable reflex. (See p. 157.) This interval (S) varied from 0.11 to 0.42 second. In 39 cases in which the measurements were fairly satisfactory, the average was 0.25 second.2 The standard deviation for our series of 39 S values is 0.076. The coefficient of variability for this refractory period measure is found to be 30 per cent.3

1Measurement No. 7 (changes in pulse-rate occasioned by short periods of exertion) is discussed with the other pulse data; see p. 415.

2Dodge and Benedict, Carnegie Inst. Wash. Pub. No. 232, 1915, p. 36.

3This average is for the reflexes of the first stimulus. On page 155 it is explained that two pendulum hammers were used to stimulate for the reflex. The second hammer was to fall against the knee at a variable interval from the stimulus of the first. When this interval was short, there was no response, that is, the reflex was refractory. When it was fairly long there was a response, in some eases almost as large as the primary reflex. These secondary reflexes, however, which followed shortly upon the primary, are not counted in the averages, σ is used as an abbreviation for 0.001 second.

We have mentioned the variability in the latency, amplitude, and the refractory period of the patellar reflex for different subjects. We might also call attention to a similar variability between successive series of records on the same subject. Dodge and Benedict have called attention to this factor and have given illustrative data on page 46 of their publication. It is known that in the case of the protective lid reflex, if we use sharp noises as stimuli for the eye wink, the individual gradually becomes accustomed to this stimulation and the wink is inhibited, either showing a large decrease in its amplitude or being absent altogether. The same series of changes occur in the case of the patellar reflex. The reason for it is not so obvious as in the protective lid reflex. Nevertheless the stimuli commonly become more or less ineffective in that the amplitude of the reflex decreases if the measurement is taken every few days. Something of the same phenomenon shows in most series of reflexes taken at one sitting. The first few of the set may be much larger than those which follow. This same condition is of course true also with lid reflexes.