98 PSYCHOLOGICAL EFFECTS OF ALCOHOL. 



comparative records of the same sounds only." The admissibility of 

 any particular type of such sound keys is first a matter of sensitivity 

 and constancy, and secondly a matter of convenience. Sensitivity of 

 the voice key affects reaction experiments chiefly through its relation 

 to instrumental constancy. The use of extremely sensitive recording 

 devices, like the phonoscope of Weiss, or the microphone, would be 

 possible, but is probably inexpedient, since, in view of the fundamental 

 defects of all records of speech-reactions by air-movement, an instru- 

 ment of such sensitivity could only give the illusion of extreme accuracy 

 in speech-reaction measurements. It would not obviate the main 

 defects of the measurement. Simultaneous records of the throat- 

 muscle movements and tested sound keys make it clear that the simi- 

 larity of sequence of the physiological processes as close as 0.001" can 

 not be relied upon even for similar sounds. The demand for an ex- 

 tremely sensitive instrument under such circumstances would be experi- 

 mental pedantry. The voice key which was used in this experiment is 

 one which was first described by Dodge. 1 Like the Erdmann-Dodge 

 key, it is a modification of the Kraepelin-Cattell sound key. The 

 present form was evolved after a considerable number of changes, to 

 make the instrument more compact, more manageable, and more 

 regular in its action. 



One end of a short brass tube, 4 cm. in diameter, is fitted with a hard- 

 rubber ring (shown removed from the brass tube in fig. 16). Across 

 the ring a rubber membrane is stretched. This membrane presses a 

 light spring, with platinum tip, against an adjustable contact-point 

 within the tube. When the spring and membrane are in elastic equil- 

 ibrium, the contact-point is adjusted by a micrometer-screw to make 

 the lightest possible contact. The contact should be tested to break 

 by a slight free-hand jerk of the key. It should break positively in 

 movements of 2 cm. Under such circumstances a slight increase of 

 air-pressure within the tube, such as is produced by speaking into its 

 open end, disturbs the elastic equilibrium of spring and membrane and 

 breaks the electric circuit. 



The relative latency of this instrument has been tested in a number 

 of ways. Records illustrating some of these tests are reproduced in 

 figures 17 to 20. All these records are read from left to right. The 

 vertical ordinates are 0.01 " apart. The horizontal ordinates are 

 approximately 1 mm. apart. 



These and similar records also give us definite controls of the total 

 latency of our voice key in series with the Harvard marker, as actually 

 used in these experiments, and also the relative latency of the Harvard 

 signal as compared with the Deprez signal. The total latency of our 

 voice key and Harvard marker is not over 2cr (0.002") for open tones. 

 The latency of the Deprez signal is not over 0.5<r. Most available 



e, An Experimental Study of Visual Fixation. Monograph Supp. of the Psychol. 

 Review, No. 35, 1907. 



