198 H. M. JOHNSON AND FRANKLIN C. PASCHAL 



3. The rate of deterioration cannot be determined accurately 

 by the slope of the latter part of the curve, as this depends in 

 large part on the relative excellence of performance before 

 deterioration began. Also, all experiments cannot be terminated 

 in the same phase of the period of work. Indeed, not infre- 

 quently, a sudden collapse occurs during a period of rest. Also, 

 the first appearance of symptoms of profound deterioration is 

 separated from the symptoms of impending collapse by a tem- 

 poral interval which varies widely among different subjects. 

 Therefore, the tune of removal is often determined by consider- 

 ations of safety of the subject, and it frequently occurs too early 

 in the work period to permit of completion of the last set of 

 material. The performance during the last period being integ- 

 rated over only a short time, the value of the last ordinate may 

 be largely determined by fortuitous factors. 



4. The deleterious effects of depletion of oxygen may be 

 compensated for, for a time, by increased voluntary effort, and 

 thus may fail of adequate representation hi the graphic record. 



With these precautions in mind the graphs, supplemented by 

 clinical notes, are useful. The principal reason for presenting 

 the results in this form is that graphic records, considered as 

 wholes, are more easily comprehended than tables. 



The writers believe that any work-test, made under the limit- 

 ing conditions which apply to the present case, is open to the 

 same objections, so that a reliable case-record is indispensable. 



The heavy ordinate drawn through the breaks of the curves 

 at the right of the graphs indicates the time of termination of 

 the experiment. The straight line drawn from abscissal zero 

 to this ordinate approximates the rate of oxygen depletion. Only 

 the initial and final percentages being given, the actual rate 

 cannot be accurately represented. The abscissal value to which 

 the curve of psychological performance is continued by a broken 

 line represents the estimated point at which complete break- 

 down would have occurred. 



