88 Prof. Carl Barus on 



the tube g ends in a hook to which the air-wire h is attached. 

 The whole arrangement is held down horizontally on the 

 sounding-board by two flat clamps (not shown) or by a ring 

 insulated from the plates a and b. Carbon powder /, is finally 

 introduced into the tube g, not in too large quantity, and not 

 packed tightly. The current therefore passes into the upper 

 plate at a, then through / into the lower plate and out at /3, a 

 and j3 being the terminals. 



My first trials with finely pulverized graphite failed. 1 

 obtained the best results with gas-carbon ground in a mortar 

 and sifted in a way to keep the particles granular. The plates 

 must be clean and bright. The sounds usually start with a 

 creak, as if the powder must first be shaken loose ; and 

 tapping frequently improves it. The plates must not be 

 pressed upon the powder, and the part of it around the stylus 

 is chiefly effective. In other forms I have quite filled up 

 the discoid cavity, but it was then frequently necessary to 

 pull up the upper plate with a spring. 



6. The following results were obtained with the micro- 

 phone of § 5. A series of wires of different diameters were 

 tested, and those larger than 0'05 cm. and smaller than 

 0*15 cm. were found best adapted for the purpose. Thicker 

 and thinner wires behaved peculiarly, as will presently be 

 seen. It has been stated that StrounaPs law requires a cor- 

 rection for thin wires ( < 0*1 cm.). The following is a record 

 of some experiments : — 



(A) Copper wire, 0*072 cm. in diameter, 190 cm. long. — 

 Sounds were heard from g' (faint), a' (faint), b', c" chromati- 

 cally to c"\ to c IV , to e lv } which corresponded to the highest 

 speeds safely attainable in my whirling arm. The range of 

 frequency is thus from n = 391 to above n — 2610 indefinitely; 

 interpreted by Strouhal's law (since d= 0*00072 met.), from 

 1*41 met./sec. to above 9*40 met./sec, or from about 6 miles 

 per hour to 21 miles per hour, and above this indefinitely. 

 The sound therefore begins with a calm or the zero of Beau- 

 fort's -scale. Strouhal's range began with frequency 840, 

 showing advantages in the microphonic method in detecting 

 low notes, some of which were indeed quite inaudible in 

 the air. 



I counted the revolutions of the whirling arm at the pitches 

 c /r , d", e h ', g /f , c f,f , e f// , g //f , and compared the data so found 

 with Strouhal's formula, obtaining values which agreed well 

 enough to evidence the correctness of pitch in both cases. 



If the notes in question could be registered automatically, 

 the limit of accuracy would be subject only to the production 

 of beats, for the sounds vary continuously ; but the ear can 



