February 2, 1922] 



NATURE 



139 



together at more or less uniform distances apart, as 

 in crystals, the orientation of individual molecules or 

 of groups of molecules being, however, arbitrary? 

 Or, on the other hand, is the spacing of the mole- 

 cules itself irregular, the solid exhibiting in a more 

 or less permanent form local fluctuations of density 

 similar to those that arise transitorily in liquids owing 

 to the movement of the molecules? The physical pro- 

 perties of amorphous solids, notably their softening 

 and viscous flow below the temperature of complete 

 fusion, would tend to support the latter view, but the 

 possibility of a closer approximation to the crystalline 

 state should not entirely be ruled out, esf)ecially in 

 view of the very interesting recent work of Lord 

 Raylcigh on the feeble double refraction exhibited 

 by fused silica (Proc. Roy. Soc, 1920, p. 284). A 

 good deal might be expected to depend on the nature 

 of the material, its mode of preparation, and heat 

 treatment. A material formed by simple fusion and re- 

 solidification of comparatively simple molecules, such 

 as silicon dioxide, might stand on a different footing 

 from a material such as ordinary glass built up by 

 chemical action and formation of complex silicates. 



If the arrangement of molecules in a vitreous bodv 

 werr irregular, the local fluctuations of optical density 

 would result in a strong scattering of a beam of light 

 passing through it, the intensity of such scattering 

 being comparable with that occurring in the liquid 

 state at the temperature of fusion of the material (see 

 note by the present writer in Nature of November 24 

 last, p. 402). On the other hand, if the arrangement 

 of the molecules approximated to the crystalline state 

 the scattering of light would be merely thaf due to 

 the thermal movements of the molecules and would 

 be much smaller. As a matter of fact, glasses exhibit 

 a very strong scattering of light, some 300 to 500 

 times as strong as in dust-free air, the Tyndall cone 

 being of a beautiful sky-blue colour and nearly, but 

 not quite, completely polarised when viewed in a 

 transverse direction. (Some glasses exhibit a green, 

 yellow, or pink fluorescence when a beam of sunlight 

 is focussed within them, and cannot be used for the 

 present purpose ; the fluorescence, even when very 

 feeble, can be detected bv the difference in colour of 

 the two images of the Tyndall cone seen through a 

 double-image prism.) Ravleigh, who observed the 

 light-scattering in glass, attributed it to inclusions, 

 some of which he assumed must be comparable in size 

 with the wave-length (Proc. Rov. Soc., iqiq, p. 476). 

 The closest scrutiny through the microscope under 

 powerful dark-ground illumination fails, however, to 

 indicate the presence of any such inclusions, and it 

 seems more reasonable to assume, in view of the fore- 

 going remarks, that the scattering is really molecular. 

 Its magnitude is of the order that might be expected 

 on the basis of a non-uniform distribution of the 

 molecules. 



Further observations with specially prepared glasses 

 and with fused silica would be of great interest to 

 investigate the influence of the chemical constitution 

 and heat treatment on the molecular texture of the 

 solid. C. V. Ramax. 



210 Bowbazaar Street, Calcutta, December 29. 



Forecasting Annual Rainfalls. 



I\ Nature of September i last, p. 12, in com- 

 menting on the remarkable rainfall at Blue Hill 

 Observatory, Mass., in July, 192 1, departures from 

 mean monthly amounts were given to show that the 

 locality was experiencing a period of excess of rain- 

 fall. Furthermore, it was intimated that the annual 

 amount for 192 1 would exceed the 3S-year normal by 

 NO. 2727, VOL. 109] 



150 mm. (590 in.). The excess actually was. 136 mm. 

 (5"35 i"-) and practically a verification. 



Many things must be considered in forecasting 

 annual rainfalls, even after the year is well advanced, 

 and no one appreciates the uncertainties better than 

 the professional forecaster. All forecasting is 

 hazardous, but weather forecasting is especially so. 



Perhaps the most important factor in estimating 

 the seasonal trend is the strength and location of the 

 North Atlantic infrabar (the "centre of action" of 

 Teisserenc de Bort). 



In "The Winds of Boston" (Harvard College 

 Observatory Annals, vol. 73, pt. 3, and vol. 83, pt. i) 

 the relation between surface-flow of the air and rain- 

 fall is given in some detail. A preponderance of east 

 and south-east winds is evident during a rainy season. 



The rainfall of 192 1 supports the views there given. 

 There was a marked deficiency of west-north-west 

 wind and a marked excess of north-east to south-east 

 wind. The average durations, 37-year normal, are : 

 West, 1708 hours; north-west, 1543 hours; and north, 

 835 hours. In 192 1 these values were decreased 

 14 per cent., 34 per cent., and 20 per cent. The 

 excess of east wind is even more noticeable. Average 

 durations are : East, 617 hours; south-east, 560 hours; 

 and north-east, 838 hours. The deficiencies were : 

 East, 57 per cent. ; south-east, 47 per cent. ; and east, 

 27 per cent. 



A just estimate of the character of a season or of 

 a year cannot be based absolutely on the quantity or 

 duration of rain. One torrential rain lasting a few 

 hours can offset weeks of rainless weather. At Blue 

 Hill, fortunately, such occurrences are rare. 



A striking illustration of the untrustworthiness 

 of rainfall as a criterion of season is found in the 

 recent flood in Texas. At Taylor, on September 9, 

 192 1, following two months of drought, there was a 

 rainfall of 587 mm. in twenty-four hours. This is the 

 greatest daily rainfall yet reported in the United 

 States following two months without any rain. This 

 city had as much rain in one day as London gets in 

 a year, yet the mean annual rainfalls of the two 

 localities are not markedly diff^erent, being 620 mm. 

 for London and about 970 mm. for the Texas city. 

 Note that the latter locality received 60 per cent, of 

 an annual rainfall in twenty-four hours. Such an 

 accidental fall must be eliminated, or any effort to 

 correlate pressure distribution, surface flow, and rain 

 becomes futile. Furthermore, it would seem that 

 efforts to link up variation in solar output and 

 seasonal conditions as determined by the amount of 

 rain are open to question. Alexander McAdie. 



Harvard University, Blue Hill Observatory, 

 Readville, Mass., January 7. 



Units in Aeronautics. 



The present writer's original letter was phrased 

 with scrupulous care to limit consideration to the 

 single proposition that consistency cannot be main- 

 tained in dynamical equations if gravity units are 

 used in expressing inertia forces, in particular not 

 by merging explicit g in the slug. An example was 

 given so conclusive as to elicit, it was hoped. Prof. 

 Bairstow's own explanation of the statement cited 

 from his text-book. 



No opinion was then offered on the validity of the 

 relation R = kpSW, or on the merits of the poundal 

 and dyne. The courtesy of this column would 

 scarcely stretch to t;he detailed statement of the 

 writer's position in respect of opinions attributed to 

 him by implication on these and other apparently 

 digressive poiiits raised bv Mr. Rowell and Sir George 

 Greenhill. ' A. R. Low. 



