October 27, 192 1] 



NATURE 



277 



focus the light from a narrow source upon the 

 film, and to allow the light which penetrates the 



beet 



'this" 



Fig. 2. 



film to fall upon a selenium cell connected with 



a battery and telephone receiver. For example, 



the arrangement shown in Fig. 3 



works satisfactorily. Here the slit S 



is illuminated by the condensed light 



from the arc, and its image is 



formed on the moving film at F. 



The light reaching the selenium 



cell, upon which, if necessary, it can Arc 



be concentrated by an additional lens, 



is caused to fluctuate as the film moves, 



so that the corresponding sounds 



are heard in the telephone. It 



is usually necessar}- to amplify the 



feeble telephone currents by means of one or more 



thermionic valves. Mr. Bergland, the Swedish 



inventor whose recent kinematograph work 



NO. 2713, VOL. 108] 



elicited this article, has apparently been able to 

 attain augmentation sufficient to actuate effec- 

 tively loud-speaking telephones so as to be audible 

 throughout a room ; but the number of valves 

 used in cascade has not been announced. 



There are certain somewhat remarkable features 

 in connection with the sound reproduction from 

 films such as those illustrated. These, although 

 already enumerated {ioc. ci't.), are worth empha- 

 sising. Strictly, in order that the sound-vibrations 

 reproduced may correspond exactly to those which 

 originally controlled the beam of light and gave 

 the photographic record, the speed of the film 

 should be the same during recording as during 

 reproducing ; but the ear and brain are apparently ' 

 capable of recognising a word even though the 

 frequencies associated with its utterance have 

 been altered in constant proportion to a consider- 

 able degree. This applies more to some words 

 than to .others, but, generally speaking, precise 

 equality of film speeds in recording and repi^o- 

 ducing is not necessary, nor does there prove to 

 be any need to take particular care to secure the 

 correct photographic density ; the articulation of 

 the reproduced sounds is wonderfully good with- 

 out this elaboration. Most remarkable of all, 

 however, is the small effect arising from widening 

 the slit image used during reproduction. It has 

 been found that this image may be made several 

 times wider than when recording without marked 

 deterioration of the reproduced words. All this 

 points to the fact that effective listening demands 

 only the existence of thie chief features of the com- 

 plicated vibrations constituting speech sounds. 

 The practical effect is that in gramophones gener- 

 ally the speed used need not be so large as to 

 record all the finer details existing in the vibra- 

 tions, and in the case of the particular optical 

 gramophone under consideration, a film speed 

 much less than that indicated above would suffice. 

 Experiment proves this to be true, and successful 

 films have been made, as was anticipated, with 

 speeds of about 40 cm. per second (the approxi- 

 mate rate in the kinematograph as ordinarily 

 used) and a slit image 0-2 mm. wide. The 

 degree of success in reproduction may be 

 judged from the fact that single words, 

 isolated from all context, are nearly always 

 recognised at once, in spite of the severe test 

 which such an arrangement obviously imp>oses. 



Moving 1 film 



FIG. 3. 



In passing from the consideration of speaking 

 films alone to their synchronous combination with 

 kinematograph pictures, we pass from facts well 



