88 
ROBERT JAMES KELLOGG 
substance or surfacing approaching absolute smoothness must 
be sought. 
Decidedly better results have been obtained by the combina- 
tion of physical levers with light-levers. Hermann's work on 
this line (cited above) is well known. It is worth while, how- 
ever, to call especial attention to Rosseks apparatus. A tiny 
concave mirror attached to the reproducing stylus of the phono- 
graph reflects a beam of light to the recording screen, thus giv- 
ing synchronous graphic and auditory records of the sound 
waves, the latter record testing and interpreting the former, and 
enabling him to perfect his apparatus. Fuller details of his 
method are given in his Recherches experimentales, cited above. 
For want of adequate funds, his experiments were confined to 
working out the principle. With a lateral magnification of 250 
times, he secured distinctive records not only of vowels, voiced 
consonants and spirants, but also of explosive t, of on and off 
glides in connected speech, and of the gradually changing form of 
.the so-called simple vowels. His apparatus is probably capable 
of being brought to an extreme degree of delicacy and perfec- 
tion, and the principle he uses may be capable of other applica- 
tions. It would be well if some of ' our American laboratories 
could help in this development. 
Sinusoidal manometric flame records. One other type of re- 
cording apparatus probably capable of being more fully per- 
fected is the sinusoidal manometric flame record of J. G. Brown. 
The direct non-sinusoidal form of the manometric flame has 
been approximately perfected by Nichols and Merritt, as noted 
above. Brown succeeded in obtaining a sinusoidal record by 
turning the exit tube of the manometric capsule on a downward 
slant, thus obtaining a curved flame with its outer portion curv- 
ing upward and describing up and down harmonic motions with 
the variations of vibratory pressure in the manometric capsule. 
By using a clear actinic flame good photographic records of the 
flame sinusoids were obtained, as shown in illustrations in the 
article just cited. By using a smoky flame a sinusoidal record 
Phys. Rev., xxxiii, 442-446. 
