312 



NA TURE 



[September 2, 1922 



rude 

 law . 



father 



mat . 

 pet . 



of each resonant circuit was variable in steps from 

 o-ooi to 2 microfarads. The inductance L was con- 

 tinuously variable from about 0-3 to 0-7 henry. The 

 resistance R was due largely to a dial box of range 1 

 to several thousand ohms, and included, in addition, 

 the resistance of the inductometer and the (perhaps 

 100 ohms) small variable coupling resistances r and 

 r' (Fig. 1). 



The nature of the numerical results is indicated in 

 Table I., which gives approximate values of the 

 frequencies and dampings of the recurrent oscillations 

 which characterise six, of the more important vowels. 

 Group frequencies (that is, voice pitches) were for 

 each vowel varied over the range 75-300 per second. 

 The first three vowels given in this table are each 

 characterised by a single train of recurrent damped 

 oscillations ; the remaining three are characterised by 

 two trains of recurrent damped oscillations. The 

 numerical values are approximate. Indeed, ccnsider- 

 able changes in the circuit adjustments in some cases 

 do not materially alter the vowel produced. The 

 problem of determining the permissible range of 

 variation for each speech-sound requires further study. 

 For the latter three vowels the relative values of r 1 

 and r., are of some importance. 

 Table I. 



Damping Oscillations. 

 Vowel. Frequency,/. Damping constant, a. 



(Unit of Time, one second.) 

 320 small (< 50) 



650 100 



1000 500 



I 750 800 \ 



' I 1500 800 I 



I 420 50) 

 ' 1 2300 50) 



cede ■ ■ ■ {2500 50} 



These results seem sufficiently interesting to re- 

 commend the apparatus of Fig. 1 to the attention 

 of students of speech-sounds. Although simple in 

 construction, this apparatus possesses considerable 

 flexibility and range. The really difficult problem 

 involved in the artificial production of speech-sounds 

 is not the making of a device which shall produce 

 sounds which, in their fundamental physical basis, 

 resemble those of speech, but in the manipulation of 

 the apparatus to imitate the manifold variations in 

 tone which are so important in securing naturalness. 



As for the disagreement between the Helmholtz- 

 Miller, or steady state theory of vowels, and the 

 Willis-Hermann-Scripture, or transient, theory, Ray- 

 leigh pointed out that the conflict was only apparent. 

 The disagreement concerns methods rather than facts. 

 Which view-point should be adopted is thus a matter 

 of convenience in a given case. When the trans- 

 mission of speech over telephone circuits is in question, 

 for example, the steady state theory often possesses 

 obvious mathematical advantages. On the other 

 hand, the quantitative data relating to the physical 

 nature of vowels which are given in Prof. Miller's 

 well-known book, " The Science of Musical Sounds," 

 expressed, as they are, in terms of the steady state 

 theory, are less compact and definite than the data 

 of Table I., which are expressed in terms of the 

 transient theory. The general agreement between 

 the two sets of data is, of course, obvious. 



The work described in this communication was 

 performed while the writer was associated with the 

 American Telephone and Telegraph Company, and 

 was carried out in the laboratories of that company 

 and of the Western Electric Company, Inc. 



John O. Stewart. 



Princeton University, Princeton, 

 New Jersev, July 8. 



Interspecific Sterility. 



Dr. Gates, in his letter which appears in Nature 

 for August 5, p. 179, emphasises the importance of 

 tetraploid species in evolution, and with this position 

 I heartily agree. Nevertheless, I do not think that 

 the difficulties in the way of free crossings amongst 

 diploid, tetraploid, and hexaploid species are so great 

 as seems at first sight probable, at any rate in the 

 Salicaceae. 



In this order Miss Kathleen B. Blackburn and 

 myself have been conducting cytological researches 

 for some time, and find the fundamental chromosome 

 number, both in Populus and Salix, to be nineteen. 

 Up to the present only diploid species have been 

 encountered in the first-named genus, but in Salix, 

 on the contrary, diploid, tetraploid, and hexaploid 

 forms have revealed themselves. In this communica- 

 tion I wish more particularly to direct attention to 

 that homogeneous assemblage known as the Capreae 

 group, which includes, in the eyes of most salicologists, 

 three genuine species, Salix caprea, S. cinerea, and 

 5. aurita — an arrangement entirely in harmony with 

 my own views. Still, so closely related are these 

 three plants that many botanists, both British and 

 continental, have refused to see in them more than 

 one polymorphic species ; similarlv, others, although 

 they admit the distinctness of S. aurita, combine 

 S. cinerea with 5. caprea. Despite this, S. aurita 

 and S. cinerea manifest themselves cytologically as 

 purely tetraploid species, while S. caprea, in the 

 main a diploid form, possesses a tetraploid race in- 

 distinguishable in the field from the commoner and 

 normal diploid type. Furthermore, what is especially 

 noteworthy, any one of these four forms can be 

 crossed readily with the other three, and the F t 

 hybrids thus obtained prove perfectly fertile inter se. 

 Not only is this the case, but, in addition, other species 

 can be brought into the chain, as, for instance, in the 

 complex cross [(Salix purpurea x S. viminahs) x S. 

 cinerea] xS. caprea (tetraploid), produced in my 

 garden, and in the still more complicated hvbrid 

 {[(5. cinerea x S. purpurea) x S. aurita] x (S. viminalis 

 x S. caprea)] x (S. viminalis x S. phvlicifolia), secured 

 by Heribert-Nilsson : in the former, two diploid and 

 two tetraploid species have taken part, and in the 

 latter, three (or two) diploid, two (or three) tetraploid, 

 and one hexaploid form. 



As a matter of fact, in the genus Salix, interspecific 

 sterility' depends, not on the chromosome complement 

 of the species concerned, but on the phvsiological 

 divergence of the groups to which they happen to 

 belong. Experiments designed to cross the diploid 

 5. tuamlra with the diploid S. purpurea turn out 

 just as fruitless as similar attempts to hybridise it 

 with the tetraploid 5. cinerea and the hexaploid 

 S. Andersoniana ; on the other hand, the hybrid 

 combinations between it and the tetraploid S. alba 

 and S. fragilis can be obtained with the utmost ease. 

 J. W. H. Harrison. 



Armstrong College, Newcastle-upon-Tvne 

 August 8. 



The Mass -spectrum of Iron. 



I have recentlv investigated this element by using 

 the vapour of its penta-carbonyl mixed with carbon 

 dioxide. It is even more troublesome to deal with 

 than the corresponding nickel compound, but bv 

 employing intense discharges and long exposures 

 fairly satisfactory results have been obtained. 



The mass-spectrum of iron is characterised by a 

 strong line, approximately at 56, and it may be con- 



NO. 2757, VOL. IIO] 



