256 CARNEGIE INSTITUTION 



records are specially made, the material is of such appropriateness 

 that it could be profitably published at once as an atlas with instruc- 

 tions for study. This would place a great amount of new data in 

 the hands of psychologists and linguists for investigation and pub- 

 lication ; or publication may be reserved until arrangements are 

 made for studying the results. 



Machine B (French curves) should also be kept running for six 

 months. During this time the machine can trace a record of French 

 prose, a record of French verse, and a record of French words, each 

 one of which, with the exception of unstudied records already in 

 my possession, will be the only accurate curves of French speech ever 

 obtained, and will furnish, for the first time, accurate data concern- 

 ing the French vowels, French melody, French rhythm, and French 

 meter. The curves could profitably be published at once as an atlas 

 of plates with instructions concerning methods of study. If pub- 

 lication were postponed. Professor Grandgent, of Harvard, would 

 presumably assign topics for doctor theses and deliver results for 

 publication with the curves. As these curves would contain ma- 

 terial for 8 or 10 theses, and as the interest is widely spread, I recom- 

 mend the prompt publication of the plates alone. 



These machines can be handled only by myself and the mechanic 

 who has had four years' experience in making and running them. 

 This mechanic (C. S. Smith, 119 Davenport avenue, New Haven, 

 Conn.) is willing to continue the work at $18 a week, or $450 for 

 the next six months. The incidental expenses — paper, paste, elec- 

 tricity, repairs, etc* — can be estimated at $75, making a total of $525. 

 These two machines will run in his charge in America. The results 

 will be delivered to me as my property, subject to the wishes of the 

 Carnegie Institution in respect to publication. 



4. Further new gramophone records should be specially made for 

 these investigations. My experience has led me to formulate the 

 following principles: 



(a) The persons selected for the records should be of such prom- 

 inence that their biographies are and will be obtainable from cyclo- 

 pedias, histories, and biographical dictionaries by any one studying 

 the results. I often find it difficult to obtain the requisite data 

 directly from persons of no distinction. All my efforts have been 

 fruitless to obtain, directly or indirectly, any information from the 

 still living speaker of the first record I studied. In any case, per- 

 sons elsewhere may study such speech curves from other points of 

 view and may wish different biographical data from the ones I have 



