1280 MELLOR: THE UNION OF HYDROGEN AND CHLORINE. CXXX.-The Union of Hydrogen and Chloyine. V. The Action of Light on Chlorine Gas. By J. W. MELLOR. IN 1843, Draper announced to the British Association Meeting at Cork (B.A. Reports, (‘Abstracts of Communications,” 1843, 9 ; Phil. Mag., 1844, [ iii], 25, l ) , ‘‘ chlorine gas which has been e x p o s a t o daylight or sunshine possesses qualities which are not possessed by chlorine which has been made in the dark. This is shown by the circumstance that chlorioe which has been exposed t o sunshine has obtained from that exposure the property of speedily uniting with hydrogen gas, a property not possessed by chlorine which has been made and kept in the dark. This quality gained by chlorine arises from its having absorbed tithonic rays* corresponding in refrangibility to the indigo.It is not a transient, but a permanent, property, the rays so absorbed becoming latent and the effect lasting an unknown time.” Draper’s experiment in support of this important statement was performed in the following manner. Equal volumes of chlorine were introduced into similar clear glass tubes and confined over salt water. The gas in one vessel was exposed to bright sunshine, while the other was kept in darkness. After a quarter of an hour’s time, an equal volume of hydrogen was added to each tube, and both tubes were then exposed t o diffuse daylight. The insolated chlorine united 6‘ promptly and energetically ” with hydrogen ; the chlorine made and kept in the dark showed no such property. * Now called “actinic.V. THE ACTION OF LIGHT ON CHLORINE GAS.1281 I n a later paper (Phil. Jhag., 1845, [iii], 2’7, 327), Draper described a modification of the above experiment in which the chlorine, dried by calcium chloride, was not allowed to come into contact with the salt water during the exposure to light, but with the same result. Draper also showed that this phenomenon is not due to the heating effect of the sun’s rays, because the chlorine will retain ics induced activity when kept some hours in the dark. I n his view, his experi- ments indicated that chlorine must be subject to a preliminary insolation before it can combine with hydrogen. In other words, chlorine after exposure to sunlight is more active chemically than chlorine made and kept in darkness.This interpretation of these experiments would furnish a simple enough explanation of the so-called period of photo- chemical induction if other investigators, working with more refined instruments, had been able to confirm Draper’s observations. The facts are as follows. In 1853, Favre and Silberrnann (Amrt. Chim. Phyls., 1853, [iii], 37, 499) quoted one experiment in which insolated chlorine was absorbed by a solution of caustic potash. They found that 478.8 calories of heat were evolved, whilst ordioary chlorine, under like conditions, gave 439.7 calories. These investigators also stated that a greater amount of chlorate was present among the products of the former reaction. On account of the complex nature of the reaction between chlorine and caustic potash, I have measured the thermal value of the simpler reaction between chlorine and potassium iodide. The gas was passed through a long spiral, kept at a constant tempera- ture, and then led directly into 500 C.C.of a decinormal solution of potassium iodide contained in a platinum calorimeter and constantly stirred by means of a platinum stirrer.’ A Beckmann thermometer was used in the calorimeter. After the temperature readicgs had been completed, the mass of chlorine which had entered the solution mas determined by titration of the free iodine with sodium thiosulphate. Otherwise I followed the details given in Thomsen’s Thermochemische Untersuchungen (1882, 2, 21, 29). The following values of R (= 100 calories) were obtained : Non-inaoIated chlorine. Insolated chlorine.283 282 287 286 281 285 I have also employed Carnot’s method (Compt. Tend., 1896,122,449) for the determination of the amounts of chlorides, chlorates, and hypo- chlorites in the products of the action of insolated and of non-insolated * I desire to take this opportunity of thanking Mr. Hartog for the loan of his VOL. LXXXI. 4 Q platinum apparatus,1282 MELLOR: THE UNION OF HYDROGEN AND CBLORINE:. chlorine on caustic potash. dekanormal solutions of potassium hydroxide at 20' :* The following results were obtained with Non-insolated chlorine. Insolated chlorine. Hypochlorite. Chlorate. Hypochlorite. Chlorate. - Y 71-58 per cent. 0.53 per cent. 70.34 per cent. 0.57 per cent. 71.72 ,, 0.47 ,, 72.01 ,, 0.51 ,, The chlorides are given by difference, total chlorine being determined as silver chloride.So far as these experiments go, the evidence in favour of the assumed existence of Draper's allotropic chlorine is not convincia g. Bunsen and Roscoe (Phil. Tram., 1857, 146, 398) failed to detect any difference between the rates of combination of hydrogen and non- insolated chlorine and of hydrogen and insolated chlorine in their extremely sensitive actinometer or to detect any action between hydrogen and insolated chlorine in the dark. t Askenasy and Meyer published the following experiment in 1892 (Annalen, 269, 72). Chlorine was exposed to bright sunlight concen- trated by means of a concave mirror and then mixed with a known volume of hydrogen in the dark. After the lapse of three or four hours, the gases were driven from the apparatus by means of a current of carbon dioxide; analysis failed to detect any difference in the quantity of hydrogen before and after contact with insolated chlorine.This experiment does not compare in delicacy with Bunsen and Roscoe's test. Traces of hydrogen could easily escape detection. Repetition. of Draper's Experimenta Two bulbs, A and B, each about 100 C.C. capacity, were connected with three cocks, a, 6, c, as shown in Fig. 1 (p. 1283). At one end, c, a capillary tube dipped into a concentrated solution of sodium chloride.$ Chlorine was led through the three-way cock b, into R and out a t D, until all the * A t temperatures above SO", the yields were very irregular. A difference of a few minutes in the time of passing the gas into the solution will make a marked differ- ence in the yield of chlorate owing to the decomposition of the hypochlorite.Fuller details will be given in another communication. j. Amato (CT'axzetta, 1884, 14, 57) states that an explosive mixture of hydrogen and chlorine will not explode in sunlight a t - 12" provided that the chlorine has not been snhjected to a preliminary insolation. No experiments are given with insolated chlorine. Butlerow and Rizza (Bull. SOC. Chim., 1882, [ii], 38, 554 ; 1883, [ii], 39, 263) appear to have also determined the atomic weights of insolated and non-insolated chlorine. $ Chlorine dissolves in, or otherwise attacks, all liquids available for use in the measuring gauge. I have always employed sulphuric acid saturated with chlorine gas unless there were special reasons, as here, for using another liquid.I have modified Draper's experiments in various ways.V. THE ACTION OF LIGHT ON CHLORINE GAS. 1283 air had been driven out of this part of the apparatus, and the salt water was saturated with the gas. The three-may cock was then closed. A was then filled with hydrogen in a similar manner with hydrogen escape at a, The cocks were lubricated with glacial phos- phoric acid. The apparatus was then exposed to bright sunIight (midsummer, 1901) for a couple of hours and then removed into a dark cellar. A and B were placed in communication for 8-10 hours by opening the cock 6. The index at first indicated that the chlorine had expanded during insolation, but in 8-10 hours the original volume was restored.There was no contraction which could be attributed to the combination of hydrogen with chlorine when allowance was made for changes of the atmospheric temperature and pressure. It will be e B 6 A a FIG. 1. evident that if the insolated chlorine had been allowed to regain atmospheric pressure by allowing some chlorine to escape before open- ing the cock 6, as, I believe, happened in many of Draper’s experiments, we should have a contraction of the expanded chlorine which has hitherto been thought to be due to the formation and subsequent absorption of hydrogen chloride by the salt water. The bulb, A, was then filled with a known volume of hydrogen at about half the atmospheric pressure ; B was filled with chlorine at atmospheric pressure. The chlorine was then exposed to sunlight and mixed with hydrogen, as before, in the dark. Next day, the mixture was swept into a burette containing an aqueous solution of caustic potash by means of a current of carbon dioxide.- Four measurements are given in the following table :1284 MELLOR: THE UNION OF HYDROGEN AND CHLORINE.Observed' 52.6 53.2 53.3 52.0 Before exposure. Reduced t o N. P. T. 50 *7 50% 50 *1 49'9 17.6" 17.8 175 17'8 Vol. at N. P. T. C.C. 771 770 777 778 50.1 50.0 50.5 49-8 Temp. 14%" 15 *1 15.3 15.0 Eight hours after exposure. Bar. mm. 773 762 770 770 - Volume C.C. Diff. c. c. + 0-6 + 0.5 - 0.4 +0*1 - I have also used the method adopted by Dixon and Peterkin (Trans., 1899, 75, 613) for observing the change in the volume of mixed gases. The inner bulb of a ball condenser was filled with hydrogen, the outer with chlorine; the mixing of the gases after exposure to sunlight was effected by breaking the inner bulb by means of a piece of glass rod previously placed inside.No evidence was obtained of any combina- tion in the dark. I am consequently led to conclude that the contraction which Draper attributed to combination of hydrogen was, in most caBes, a secondary effect of what may be called the photo-expansion of chlorine (the Budde effect), namely, the contraction which insolated chlorine always undergoes on removing the light, The Bzldde Efect. I n 1870, Budde (Phil. Mag., 1871, [v], 42, 290; Pogg. Ann. Ergbd., 1873, 6, 477) discovered the fact that chlorine expands under the influence of light rays of high refrangibility. The temperature of the chlorine rises about one degree.This is not a direct heating effect of sunlight, because the interposition of a screen of water between the source of light and the chlorine makes no difference to the effect. Favre and Silbermann (Zoc. cit., p. 499, footnote), so far back as 1853, tried the same experiment as Budde, but with a negative result. As a matter of fact, no change in volume does occur if the chlorine is thoroughly dried (Roscoe, Watts' Dictionary, 1875, 7, 750 ; Baker B.A. Reports, 1804, 493; Shenstone, Trans., 1897, 71, 471). Richardson, however, has not only established the reality of the Budde effect (Phil. Mag., 1891, [v], 32, 277), but he has also shown that the magnitude of the photo-expansion is proportional to the intensity of the more refrangible rays of light, Hence he utilises this property to measure the intensity of rays of high refrangibility.TheV. THE ACTION OF LIGHT ON CHLORINE GAS. 1285 expansion is independent of temperature between the limits 14’ and 1 3 8 O . I f I denotes the intensity of the light and ZI the volume of the chlorine, I = av, where a is a constant very nearly equal to 70. The expansion occurs when chlorine is mixed with air (Richardson), or with hydrogen, carbon monoxide, or ethylene (Recklinghausen, Zed. physikaZ. Chern., 1894, 14, 494). (‘ Ausdehnung durch die Reak- tion,” says Recklinghausen, ‘‘ scheint demnach das Charakteristikum einer durch das Licht stattfindenden Vereinigung zweier Gase zu sein.” Recklinghausen also measured the magnitude of the photo-expansion of chlorine mixed with hydrogen, not directly in contact with water, in an apparatus in which the mechanism of a plethysmogreph is ingeniously applied to measure slight changes in volume.He found that as the chlorine disappeared, owing to the formation of hydrogen chloride, the mixed gases returned to their original volume. When the light was removed a t any stage of the exposure, the mixture also returned to its original volume. If the light were again restored, the mixture again expanded, but not quite to the volume occupied just before the light was removed. From this, Recklinghausen argues that it is possible combination takes place in the dark. Another interpretation may, however, be suggested. The total increase in the volume of the actively combining gases is the additive effect of (1) Budde’s photo-expansioo, and (2) the heat of combination of hydrogen and chlorine gases.I n darkness, combination ceases, the mixture cools. When light is restored, the resulting expansion is the joint effect of the photo-expansion of the remaining chlorine and the heat of combination of the newly formed hydrogen chloride without the thermal effect which accompanied the formation of the hydrogen chloride produced during the first exposure. Budde has suggested the following possible explanations of his dis- covery : (1) Light loosens or decomposes the chlorine molecule into free atoms, so that c1, = c1 + c1. Two volumes. Four volumes. (2) Light performs some work which changes into heat, thereby oaus- ing an expansion.(3) The chlorine is warmed up directly in the same way that lamp black is warmed up in the heat spectrum. He rejected the second of these suppositions as a makeshift (Nathbehelf) ; the third he rejected because the heat rays at the red end of the spectrum pro- duce little, if any, photo-expansion. The first supposition appeared to1286 MELLOR: THE UNION OF HYDROGEN AND CHLORINE. explain most of the facts, and he suggested that the slight rise of temperature was developed by the recombination of the ‘‘ gelockerten und zersetzen Clhlormolekule.” The first question I have attempted to answer is : Does the rise in temperature account for the expansion 1 Professor Dixon has suggested to me the following instrument. The two concentric bulbs of a ball condenser (Fig.2) were respectively fused on to two sulphuric acid gauges, The outer bulb was filled with chlorine, the inner with air, so as to form an air thermometer sur- rounded by the chlorine. The stem of each bulb was graduated by warming up the bulbs in a suitable vessel and marking the position of the index in each stem a t the different temperatures. The two indices show practically the same rise of tempera- ture when the bulbs are placed in a water-jacket and bathed in sunlight. The sir thermometer probably shows more correctly the temperature of the surrounding chlorine than a mercurial thermometer. FIG. 2. The photo-expansion is therefore proportional to the temperature of the gas, and it is unnecessary to as- sume that the chlorine molecule is dissociated when exposed t o actinic light. Measurements of the conductivity of the illuminated gases confirm this conclusion. J.J. Thomson (Proc. Camb. Phil. SOC., 1901, 11, 90; see also Hemptinne, Zeit. physikal. Chem., 1902, 12, 244) has experimented on this subject. He says: “The object of these experiments was (1) to see whether there were any free ions present either in the preliminary stage when the expansion discovered by Draper is oc- curring, or (2) when the hydrogen chloride is being produced. At neither stage could I detect any free ions, although I could have done SO had the ions amounted to anything like 1 in 1014 of the molecules present. I then tried whether the rate of combination was affected when ions were produced by external means, for example, Rontgen rays, thorium radiation, &c.I could not detect the slightest effect.” According to Clausius’ theory, for any given temperature there i s ~l definite relatios The results were negative. To return to the hypothesis rejected by Budde.v. THE ACTION OF LIGHT ON CELORINE GAS. 1287 between the molecular and the atomic energy of a gas, or between the rectilinear motions of translation and the vibrations of the molecules (or atoms). The former motions are supposed to determine the t e a - pernture of a gas as measured by a thermometer. If the molecular vibrations are augmented, the increased energy is converted into energy of translatory motion either after a few or after a great number” of molecular impacts. If the molecular vibrations are con- tinually excited by an impressed force, such as the absorption of light rays of a definite period of vibration, there will be a continuous trans- formation of this energy into energy of translatory motion, which makes itself evident by a rise of temperature.If the internal energy is not immediately dissipated by, say, radia- tions from the molecules moving in their free path between successive collisions, or by the conversion into heat during molecular impact, energy will be stored in the gas and ultimately make its appearance as luminescent or phosphorescent phenomena (Wiedemann, Ann. Phys. Chim., 1889, [ii], 37, 177; also Phil. Uag., 1889, [v], 28, 149, 248, 376, 493). I find that no radiations capable of affecting the fastest ‘‘ Paget ” sen- sitive plates? are evolved by chlorine gas contained in thin glass tubes exposed for 1-5 hours to bright sunlight, limelight, or to a 600-1000 c.p.electric arc ; nor does such a vessel of chlorine emit any rays capable of causing a perceptible influence on a sensitive mixture of hydrogen and chlorine, A layer of 15 cm, of moist chlorine appar- ently affords perfect protection from the actinic rays capable of effect- ing the union of hydrogen and chlorine. The actinic energy from sunlight is apparently absorbed and subsequently dissipated in the form of non-actinic external radiations. If the amount of expansion of isolated chlorine could be taken as a measure of the work performed by the absorbed actinic energy, a simple calculation would furnish the thermal equivalent of any ray of light absorbed by the chlorine.If the gas obeys van der Waals’ law, for example, the work of isothermal expansion is where a and 6 are van der Waals’ constants, R is the gas constant, 8 the absolute temperature, vl the initial, v2 the final, volume of the gas. It is clear that if the ‘‘ kinetic potential ” of the energy changes * Stoney’s “88” and <‘Bb’’ events respectively (Stoney, PhG. Hag., 1895, + I desire to thank Mr. Bradley for doing the photographic work. Evl, 40, 262).1288 MELLOR: TEE UNION OT HYDROaEN AND CHLORINE. during the process of absorption-of light be studied thermodynamically, the entropy must, a t the outset, be divided into two parts. The one part corresponds with temperature as defined by the motion of the translation of the molecules which is indicated on the thermometer ; the other part corresponds with temperature ” as defined by the mole- cular (or atomic) vibrations and is not indicated on the thermometer.I have tried to find further indications of the nature of the internal work performed when light is absorbed by chlorine gas by comparing the ratio of the two specific heats for chlorine in sunlight, and chlorine in darkness, at the same temperature. I was unable to detect any difference outside the range of experimental error with an ordinary Kundt’s tube. This only shows that the change in the internal energy of the isolated gas is an immeasurably small fraction of the total energy of the translatory motions of the molecules. Chemical Action of Light on Moist Chlorine. Richardson (B.A. Reports, 1888, 89) has mixed moist carbon dioxide with sufficient chlorine theoretically to decompose all the moisture, and found that 1.33 per cent.of the chlorine had been con- verted into hydrogen choride after 33 days’ exposure to light ; when oxygen was substituted for the carbon dioxide, 3 per cent. of the chlorine was converted into hydrogen chloride in the same time. Pedler (Trans,, 1890, 57, 613) has investigated the action of tropical sunlight on chlorine in the presence of liquid water in sealed tubes. Hs found that little action occurred unless the water is in relatively large excess. Moist chlorine in the beam of a 600-1000 c.p. arc light concen- trated by a lens (about 12 in. focus) shows a faint cloudiness re- sembling the cloud observed by Tyndall (Tyndall’s ‘‘ Heat, a Mode of Motion,” 1880, 475) with sulphur dioxide and the vapours of certain organic liquids, I have separated a little oxygen* (about 4 per cent.of the total volume) by passing the illuminated gas from Tyndall’s tube, a t a temperature of 20-25”,l. into a cold solution of potassium hydroxide by means of a current of carbon dioxide. The oxygen was not due t o the action of chlorine on any visible moisture deposited on the illuminated parts of the apparatus. Morren (Ann. Chirn. Phys., 1870, [iv], 21,323) found that hydrogen chloride, similarly exposed to a beam of sunlight, was not decomposed. * It is extremely difficult to prepare large quantities of chlorine free from traces of air. By means of the process indicated in Part I, it is possible to prepare 10-14 litres of chlorine contaminated with immeasurably small amounts of air. Thig method of preparation was employed here.t. The gas in the vicinity of the focus of the lens is no doubt at a higher tempera. ture than this.V. THE ACTION OF LIGHT ON CHLORINE GAS. 1289 (i) Air in ,,. 2 1 3 a 13 12 14 16 18 18 If chlorine is dried by means of phosphoric oxide, there is no sign of Budde’s expansion in sunlight. (Concentrated sulphuric acid may be used as index liquid.) Consequently, the amount of actinic energy which is converted by dry chlorine into heat is not sufficient t o affect the chlorine thermometer above described. Cordier (Monatsh., 1900, 21, 660) finds that whilst moist chlorine is opaque, dry chlorine is more transparent than moist chlorine t o actinic light so far as the screening effect of chlorine on silver chloride is concerned. I have fitted up the insolation vessel of Bunsen and Roscoe’s actino- meter A B C (Fig, 3) in the middle of a large globe, D, filled with FIG.3. - (ii) Dry chlorine (iii) Moist chlorine in D. in D. 0 0 0 0 0 No action 0 observed. 0 0 1 2 chlorine gas. The large globe was painted in such a way t h a t the insolation vessel could be illuminated by a beam of light which had been filtered through 10-15 cm, of chlorine gas. The following numbers are representative of the readings : Time in miuutes. a 9 10 11 12 13 14 15 16 2012%) MELLOR: THE UNION OF HYDROGEN AND CHLORINE. The dry chlorine had been in contact with purified phosphoric oxide for five months.These results raise the question : Is the actinic energy absorbed by moist chlorine gas spent in doing chemical work? The reaction between chlorine gas and water vapour is endothermic (compare Namais, Gaxxettcc, 1896, 26, i, 35). The filtering of light from the actinic rays by moist chlorine is a continuous process. When a bulb of hydrogen and chlorine gas is placed in a sufficiently large jar of moist chlorine gas and exposed to light, the bulb is perfectly screened from the actinic rays for an indefinite period. If, therefore, the absorption of light induces the above reaction, there must be a point at which the reverse reaction takes place, and is reversible, Richardson (B.A. Reports, 1888, 89 ; 1889, 59 ; 1890, 263 ; Trans., 1887,51, 801) has demonstrated that hydrogen chloride in the presence of excess of oxygen is almost completely decomposed according to the reaction 4HCl + 0, = 201, + 2H,O.Harker (Zeit. physikal. Chem., 1892, 9, 673) has also established the reversibility of the reaction at high temperatures. There is a curious relation between the equilibrium constant and the intensity of light which remains to be investigated. If a mixture of equal volumes of oxygen, hydrogen, and chlorine be exposed to actinic light, the colour of the chlorine at first disappears owing to the relatively fast irreversible reaction 2H,O + 2C1, 0, + 4HC1, H, + C1, = 2HC1, this being followed by the slow reversible change, 4HC1 + 0, ZZ 2H,O + 2C1,, the characteristic green colour of chlorine slowly returns. Since the volume of moist chlorine is alone affected by the absorption of actinic energy, the increased temperature of insolated chlorine appears to be an effect of some chemical reaction between water vapour and chlorine gas.* The amount of moisture retained by a gas which has been in contact with any given desiccating agent a sufficient length of time for thorough desiccation is of vital interest.Dittmar and Henderson * E. R. Laird (Astrophysical Journal, 1901, 14, 85) has recently mapped the absorption spectrum of ordinary chlorine. A comparison of this with that of pure chlorine, thoroughly dried by means of specially purified phosphoric oxide, would up doubt furnish useful information on this subject,v. THE ACTION OF LIGHT ON CHLORINE GAS. 1291 (Proc. Roy. Xoc. Glasgow, 1891, 22, 33) found that air dried by calcium chloride may retain 0.001 gram of moisture per litre.Morley (Amer. J. Sci., 1885, [iii], 30, 140) found that the absolute amount of moisture retained by air dried by sulphuric acid (sp. gr. 1.8381) is 0.000002 gram per litre. Dibbits (Zeit. anal. Chem., 1876, 4, 180) found that air dried by sulphuric acid gave up 0*000008 gram per litre to phos- phoric oxide. If the degrees of accuracy of the two latter measurements permit a comparison, it follows that the absolute amount of moisture retained in a gas thoroughly dried by phosphoric oxide is immeasurably smaEZ. [NOTE ADDEDAUGUST 19 :-Mr, H. Brereton Baker, F.R.S., hasrecently discovered that “dry chlorine attacks platinum in light but not in darkness.” (Private communication.) It is thus to be expected that the heat of this chemical action will cause an expansion which, if sufficiently great, will more than counterbalance the contraction due to the absorption of chlorine in the formation of platinurn chloride.A t a meeting of the Chemical Society, in November 1900, in the discussion of an earlier part of this paper, Mr. Baker mentioned the fact that dry chlorine in the presence of platinum expands in sunlight.] Note on the Action of the EZectric Discharge on Moist Chlorine. According to Henry (Phil. D*ans., 1800, 190, 238), 1/35 volume of hydrogen chloride is decomposed by the passage of an electric spark through the gas, Vernon (Chem. News, 1891, 63, 67) found that the passage of a silent discharge through chlorine in an Andrews’ ozone tube resultsin a slight expansion due to the heating effect of the discharge.I have mixed hydrogen chloride and’ chlorine and repeated these experiments to try if ‘‘ nascent ” chlorine from the hydrogen chloride would form anything of tohe nature of C1, or HCl,, but without success.* If, however, air or oxygen be present, the ozone formed will react with the hydrogen chloride. Conclusions. 1. There is no experimental evidence to show that chlorine gas under the influence of light undergoes an? change capable of appreci- &ly affecting the chemical activity of that element towards hydrogen. 2. Part of the energy absorbed by moist chlorine from sunlight is dissipated as heat. This causes the Budde effect. * If a tube with platinum or platinum-iridium electrodes is used, there is invari- ably a contraction in the volume of the mixed gases due to the combination of chlorine with the metal of the electrodes. I have never met with a specimen of platinum capable of withstanding the action of chlorine for many days without the fQrmation of a reddish “rust ” of platinum chloride over its .wrface.1292 MELLOR: THE UNlON OF HYDROGEN AND CHLORINE. 3, Under the influence of light, chlorine sets up and maintains in a state of equilibrium a reversible reaction with water vapour, possibly 2H,O -I- 2C1, 4HC1 -I- 0,. 4. Dry chlorine does not exhibit the Budde effect. 5, The rise in the temperature of imperfectly dried chlorine when exposed to sunlight appears to be due to some chemical reaction between the moisture and the chlorine gas. 6. A layer of moist chlorine just thick enough to screen a bulb of mixed hydrogen and chlorine gases from chemical action is not sufficient to prevent chemical action if the chlorine is dried by means of purified phosphoric oxide. 7, The actinic energy continuously absorbed from sunlight by moist chlorine is continuously dissipated in at least three ways : (i) partly in maintaining the above chemical reaction ; (ii) partly by conversion into heat during molecular impacts ; (iii) partly as external non-actinic radiations from the molecules moving in their free path between mole- cular collisions. THE OWENS COLLEGE, MANC'HESTER.