3 68 



NATURE 



\ i. 1919 



ition occurring until the density of the vapour 

 was about eight times that of saturation al the 

 tempi iture, but that beyond this point the 

 condensation was so dense as to suggest that a 

 1 atural limit of supersaturation had been re; 

 Experiments on steam-jets by Barus and others 

 suggested a similar limit for steam at high tcm- 

 ures free from air, though the precise ratio 

 ol density required could not be directly obtained 

 from such experiments. Taking the densit) ratio 

 given by Wilson as the limit of supersaturat on, 

 the discharge through a nozzle was calculated bj 

 Callendar (Proc. tnsl . January, 1915), 



and shown to afford a reasonable explanation of 

 well-known anomalies. 



Admitting' the supersaturation limit thus defineS 

 the discharge through a nozzle comes out about 

 5 per cent, greater than that given by the older 

 view, and agrees much better with the results of 

 experiment. The available heat-drop to the super- 

 saturation limit is about 5 per cent, less, involving 

 a corresponding loss of .work. So far the result 

 is definite, depending only on the limit assumed 

 and the equation of the adiabatic, which is fairly 

 certain. Beyond this point the loss must depend 

 on the rate of expansion, but it is still possible 

 to calculate an upper and a lower limit. The 

 maximum heat-drop is obtained by assuming that, 

 when once condensation has started, the tempera- 

 ture follows the ordinary saturation limit in isen- 

 tropic expansion, in which case there is no further 

 loss of available heat-drop. On the other hand, 

 assuming- that the temperature cannot fall appre- 

 ciably below- the supersaturation limit, however 

 rapid the expansion, the maximum loss is obtained 

 bv assuming that the temperature follows the 

 supersaturation limit, in which case the loss con- 

 tinually increases with increase of entropy, but 

 reaches a nearly constant percentage, about S per 

 cent., of the total heat-drop at low pressures. 



In the work before us the author adopts a 

 slightly different definition of the supersaturation 

 limit. Instead of taking a simple ratio of densities 

 as proposed by Wilson, he assumes that the 

 effective radius of the supersaturation nucleus 

 remains constant at different temperatures. In the 

 absence of experimental evidence at high tempera- 

 tures, it is scarcely possible to decide between the 

 two assumptions, except that the first is the 

 simpler in application. The two corresponding 

 curves for the supersaturation limit agree so closely 

 at pressures between 1 lb. and 15 lb. that they 

 give practically identical results when applied to 

 any turbine problem. Now that one of the leading- 

 exponents of turbine theory has set the example 

 we may confidently expect that other useful appli- 

 ( ations of the supersaturation hypothesis will 

 follow, and that more accurate determinations of 

 the limit will be made in the near future. 



The "New 'I hours " gives an example of one 

 such application of great practical interest — 

 namely, the effect of superheat in improving- the 

 which 1 ins the hypothesis of super- 



saturation, and throws light on the probable state 

 of the steam in an actual turbine by comparing 

 theory with experiment. 



NO. 2567, VOL. 102" 



Ai-i ording to the older theory of isentropic ex- 

 pansion of saturated steam, the effect of a mode- 

 gra ol iuperheat in improving the relative 

 ■icy of a turbine should be practically negli- 

 gible, whereas even the earliest experiments in 

 this direction showed that the improvement was 

 strongly marked. The improvement was g< nerally 

 attributed to elimination of friction due to the 

 presence of water (Stodola, "Steam Turbines," 

 p. 137), but Osborne Reynolds showed this ex- 

 planation to be unsatisfactory. The supersatura- 

 tion theory of expansion requires that the improve- 

 ment should be most marked in the early stages 

 of superheat, owing to the reduction of supersatu- 

 ration losses, which diminish most rapidly with 

 the first rise of temperature. The most trustworthy 

 and recent experimental results on the improve- 

 ment due to superheat are probably those given 

 bv the correction curves of Baumann (Journ. Inst. 

 Elect. Eng., 191 1 ), which are generally regarded 

 as accurately representing the case of the modern 

 high-speed turbine. 



.Mr. Martin show-s that these results can be 

 satisfactorily explained on the supersaturation 

 hypothesis provided that we are prepared to admii 

 that the temperature of the steam, after condensa- 

 tion has set in, remains much nearer the super- 

 saturation than the saturation limit, dividing the 

 interval in the ratio 1 to 4. His method, involv- 

 ing the estimation of reheat factors, may appear 

 indirect, but tends, if anything, to exaggerate the 

 effect of superheat in reducing the supersaturation 

 losses. The present writer has made many 

 similar calculations, which corroborate Mr. 

 Martin's, and tend to show that the temperature 

 of the steam must be very near the supersaturation 

 limit in the later stages of expansion in a high- 

 speed turbine. A result so strongly at variance 

 with the generally accepted theory cannot fail to 

 act as a stimulus to further research on the effects 

 of supersaturation, and may lead to appreciable 

 improvements in design when proper account is 

 taken of these essential physical properties of 

 steam H. L. Callendar. 



THE EPIDEMIOLOGY OF PHTHISIS. 1 



TUBERCULOSIS, and particularly pulmonary 

 tuberculosis (phthisis, or consumption of 

 the lungs), still remains one of the health problems 

 of the age. The Medical -Research Committee 

 has, therefore, been well advised to institute an 

 inquiry into the epidemiology of phthisis in Great 

 Britain and Ireland, of which this report by Dr. 

 John Brownlee is the outcome. 



The present investigation is a statistical analysis 

 of the Registrar-General's returns of mortality, 

 mainly for the five decades 1851-60 to 1891-1900, 

 for t'ne constituent countries as a whole and 

 also for certain districts of them. By this means 

 remarkable differences are brought out respecting 

 the age at which the maximum death-rate from 

 phthisis occurs in different localities. If we take 

 the deaths of males from phthisis in England and 



1 An Investigation into the Ep 

 anil Ireland. Medical Research Committee, Special Report 



