September i, 1923] 



NATURE 



323 



Letters to the Editor. 



\The Editor .. does not hold himself responsible for 

 opinions expressed by his correspondents. Neither 

 can he undertake to return, nor to correspond with 

 the writers of rejected manuscripts intended for 

 this or any other part of NATURE. No notice is 

 taken of anonymotes communications.^ 



Photochemical Production of Formaldehyde. 



In a recent paper (J. Amer. Chem. Soc, 45, 1184 

 (1923)) Spoehr states that he has been entirely unable 

 to reproduce our results on the photochemical pro- 

 duction of formaldehyde from carbon dioxide and 

 water (Trans. Chem. Soc, 119, 1025 (1921)), and he 

 invites us to describe our experimental methods in 

 greater detail than we did in our original communica- 

 tion. Before replying to this invitation we have 

 repeated all our experiments and the new results 

 entirely confirm the old, and there seems, therefore, 

 to be little or no doubt that by the action of short 

 wave ultra-violet light on aqueous solutions of 

 carbonic acid formaldehyde is formed. 



The quartz mercury lamps employed in all our 

 investigations are the U form made by the Hewittic 

 Company and the current taken by each lamp is 

 3-5 amps, at 230 volts. About 75 c.c. of pure con- 

 ductivity water were placed in a transparent quartz 

 test tube 9x1 in., and a slow stream of carbon 

 dioxide, prepared from pure marble and synthetic 

 hydrochloric acid and washed with a solution of 

 potassium bicarbonate, was passed through the water. 

 The carbonic acid solution was kept cool by a narrow^ 

 tube through which a stream of cold water was passed. 

 Stringent precautions were taken to guard against 

 contamination by organic matter, and the times of 

 exposure varied from 18 to 72 hours. Since the 

 most satisfactory test for formaldehyde appears to 

 be Schryver's test we have used it in every case, and 

 throughout the whole series of observations' we carried 

 out control blank experiments. The results obtained 

 may be summarised as follows : — 



1. No formaldehyde can be detected in the solutions 

 if the distance between the lamp and test tubes is 

 less than six inches and no screen is interposed. 



2. Formaldehyde can be detected in the solutions 

 if the distance between the lamp and test tubes is 

 six inches or more. 



3. The (quantity of formaldehj^de formed is in- 

 creased if a plate of calcite is interposed, and in this 

 case the distance between the lamp and test tubes 

 can be reduced with advantage. 



4. The quantity of formaldehyde found is increased 

 if the solution contains calcium or potassium bicar- 

 bonate. 



The amount of formaldehyde found, though ab- 

 solutely definite, is \er\ small (i to 2 parts in 100,000), 

 the reason for this being twofold. The absorption 

 band of carbonic acid lies near to X— 220MM and the 

 intensity of the radiation of the mercury lamp at or 

 about this wave-length is exceedingly small, so that 

 the velocity of formation of formaldehyde must 

 necessarily be very slow, even assuming' that the 

 whole of the radiation is absorbed. A second factor 

 is that formaldehyde in dilute aqueous solution is 

 decomposed by very short wave-length light. Indeed, 

 a o-oi per cent, solution of formaldehyde through 

 which carbon dioxide is passed is entirely decomposed 

 in 24 hours if placed at a distance of 4 inches from the 

 quartz mercury lamp. It follows, therefore, that the 

 formaldehyde "found in the solutions described above 

 is only the excess of that formed over that decom- 

 posed. The very short wave-length radiations are 

 more absorbed by air than is light of wave-length 



NO. 2809, VOL. 112] 



220yLi/u, and thus an explanation is found of the fact 

 that a minimum distance between reaction vessel and 

 quartz lamp is necessary for positive evidence to be 

 obtained of the production of formaldehyde. Since 

 calcite absorbs all rays of shorter wave-length than 

 2i^fj./j., the amount of formaldehyde is increased if a 

 calcite screen is interposed and the minimum distance 

 between lamp and test tube becomes no longer 

 necessary. 



In view of our two series of positive results it is 

 surprising that Spoehr finds himself unable to confirm 

 this reaction, since the evidence we have obtained 

 seems to us to be conclusive. In his paper Spoehr 

 states that he used the straight form of mercury 

 lamp, and in this may be found a possible explanation 

 of the failure which he has recorded. It is a known 

 fact that the quartz mercury lamp deteriorates after 

 use and loses its power of radiating short wave ultra- 

 violet light. Prof. Allmand has proved this de- 

 terioration of a mercury lamp of the straight form 

 and has been kind enough to communicate his results 

 to us. It may be suggested that this fact explains 

 Spoehr's failure to observe any formaldehyde, the 

 necessary ultra-violet radiation from his lamps being 

 too small in amount. We have noted that the 

 U-shaped lamp does not deteriorate, or, if so, very 

 slowly, since our lamps after many months' con- 

 tinuous use still ozonise the oxygen of the surrounding 

 air, a photochemical reaction which is known to be 

 stimulated by very short wave-length light (X = 2oo/j-/j.). 



The great dilution of the formaldehyde necessitates 

 the use of a colorimetric test for its detection. In 

 view- of Willstatter's statement that the Schryver test 

 is given only by formaldehyde and hexylenic aldehyde, 

 this reaction lias commonly been accepted as positive 

 evidence for formaldehyde. We have, therefore, em- 

 ployed this test, having at the same time proved for 

 our own satisfaction that it certainly is capable of 

 detecting formaldehyde at concentrations of i in 

 1,000,000. ' E. C. C. Baly. 



I. M. Heilbron. 

 W. F. Barker. 



Correlation of Upper Air Variables, 



In view of the importance of the subject, a few 

 remarks with regard to the note in Nature of May 

 19, p. 684, on " Correlation of Upper Air Variables" 

 may perhaps be permitted me, chiefly with the 

 object of making clear the real issues in this question. 

 Dines ^ found very high coefficients of correlation 

 (of the order of o-8) between various upper air 

 variables, specially with pressure at 9-kilometre level. 

 This led to the formulation of the Dines-Shaw theory 

 of the sub-stratosphere and the regions above 9 

 kilometres as the real seat of origin of meteorological 

 causes. In 1920, Chapman ' applied certain statistical 

 corrections to the coefficients of correlation found 

 by Dines and raised these to -f-ioo in several in- 

 stances. A correlation of -}-i-oo establishes absolute 

 causal nexus. A conclusion of this nature demands 

 close scrutiny, specially as it is being widely quoted 

 and applied in current writings.' In a recent memoir * 

 noticed in Nature * I have examined the statistical 

 analysis in some detail. 



As regards Chapman's work, my chief criticism is 

 this : he has neglected entirely the effect of correla- 

 tion between " errors " of measurement. Taking 



' M.O. No. 2106, Geopliys. Mem. 2, 1912 ; M.O. No. 220c, Geophys. 

 Mem. 13, 1919, etc. 



• Proc. Roy. Soc. 98 A (1920), pp. 235-248. 



• M.O. No. 22oi, Geophys. Mem. 19, p. 215 ; Sir Napier .Shaw, " The 

 Birth and Death of Cyclones." 



• Mem. Ind. Met. Dcpt., vol. xxiv. Part ii., " On Errors of Observation 

 and Upper Air Relationships." 



• Nature, May 19, 1923, p. 684. 



