It may be noted that the blue cloud (6) which appears round an electrical machine has been shown to be due probably to hydrogen peroxide and not to oxides of nitrogen, although the latter may also be present (19). In certain circumstances, however, oxides of nitrogen can cause condensation with the resultant formation of a cloud.

The dense clouds formed when hydrogen chloride and ammonia interact have been studied a great deal also. In this case practically all the work has been concerned with the sizes of the particles or their electrical charges. Here we have the possibility of two kinds of fog, “dry” and “moist” according as the gases are dry or saturated with water vapour, before the reaction; they may also be acid or alkaline according as excess of acid or alkaline vapour is used. It is to be expected that there will be a difference in the properties of the particles according to the type of fog obtained, and this expectation is borne out as far as “dry and “moist” neutral particles are concerned, especially in regard to case of absorption in liquids and to the sizes of the particles; the “dry” fogs being absorbed better in salt solutions than the “moist” ones although with pure water the opposite is the case (23, 24, 25). For acid and alkaline fogs very little data are available. It is very unfortunate also that in all the investigations known to the present author in which these fogs or clouds have been examined, no analyses have been made

Volatile organic bases “fume” strongly with hydrogen chloride vapour, e.g. pyridine, picoline, etc.; all give intense clouds when they or their solutions are brought near the acid vapour (29).

With sulphur trioxide there is also the possibility of two types of particle, “dry” or “moist.” In this case the “moist” fog will really be a concentrated solution of sulphuric acid. The “dry” are well known technically and the problem of absorbing them completely is very important. Similarly the “moist” fogs are important being often troublesome in sulphuric acid concentration plants (13, 30). Sulphur trioxide in the vapour state on being passed through water condenses to form a dense fog (7). The problem of the stability and absorption of these fogs will be considered later. The same remark as was made for ammonium chloride has to be made concerning these fogs; no case has been found in the literature in which particle size and concentration of acid in the particles have been given together, and only one case has been found in which the composition and temperature of the vapour were given with the concentration of acid in the particles (see 13, 21, 23, 24, 25, 30).

Hydrochloric acid and nitric acid mists are also known technically. The former is particularly important because of the use of easily hydrolysable chlorides for cloud formation in chemical warfare (12). Many of the best cloud-producing agents consist of such a chloride base. These acid “fumes” are well known in laboratory work also, and in some cases have caused slight difficulties in atomic weight determinations. Many determinations have been made of the particle size and electrical charge of these fogs, but again no analyses are available. When the “fumes” from concentrated

hydrochloric acid are examined it is found that the radius of the particles is about 5 × 10^-5 cm. (20), i.e., practically the same as that found by Rothmund (l.c.) for his ozone fogs. But the formula which Rothmund uses cannot apply to these fogs as the particles must consist of concentrated acid. Even with the comparatively dilute solutions of acid in the fog particles obtained in experiments of the present author, this formula does not appear to be of any use, and here the particles are much larger, namely, 2 × 10^-4 cm. radius. Hydrobromic acid behaves in a similar but somewhat more intensified manner to hydrochloric. Its solutions “fume” more strongly; the bromides of weak bases are also easily hydrolyzed with production of a fog (5).

It is well known that the amount of “fuming” is dependent on the water vapour partial pressure of the gas in contact with the acid or the chlorides (18). An attempt has been made to get quantitative data on this point, but, although the quantity of weighable “fume” appeared to increase as the percentage humidity increased (the temperature and partial pressure of the hydrogen chloride being constant), the results were not sufficiently definite to warrant their being included here.

There are numerous references in the literature to the production of clouds in the presence of radium or of ultra-violet light (4, 9). It is not necessary for water to be present, as the vapours of many organic liquids will also condense under these conditions either in air or other gases such as hydrogen and carbon dioxide. In these cases there is a considerable amount of evidence to show that the presence of ions is not always the controlling factor (1), indeed Curie states (9) that the fogs formed in the presence of radium A have no connection with the condensation clouds with ions. Curie found that the presence of compounds, sometimes due to the use of rubber stoppers, absorbable by water caused the fogs to be much more intense. And Jones in 1925 (16), concluded that on illumination of moist halogen-air mixtures fogs were only formed when an oxidizable impurity was present.

Townsend (28) has examined the fogs formed during electrolysis. His results are discussed in a later section of this summary in conjunction with the results of others (8) on the production of electrical charges in gases by bubbling through liquids.

It is interesting to examine the properties of atmospheric fogs in relation to the explanations which have been put forward to explain their stability in unsaturated atmospheres. Because of this stability Frankland (11) has called such fogs “dry fogs.” But his explanation, namely, the presence of an oil film preventing evaporation cannot be considered correct. Neither can the lowering of vapour pressure due to dissolved materials be sufficient to allow the particles to exist in a very unsaturated atmosphere. Owing to the comparatively large diameter, 1.4-3.5 × 10^-3 cm., of these particles the curvature of the surface can have only a very small effect on the vapour pressure.

Investigations have shown that even with so-called “smoke fogs,” that is, fogs which are made intense owing to smoke and acid fumes, the humidity of the air may be anything from 76 to 100 per-

has also drawn attention to this point in regard to the efficiency of gas masks.

Tyndall found during his experiments on optically pure air that dust motes of certain types were extremely difficult to remove; in many cases removal was possible by prolonged contact with concentrated sulphuric acid, the explanation being that the acid removed the adsorbed gas film and probably even brought about the complete destruction of the particles (15, 16). The question of the removal of solid particles also played an important part in the development of the experimental work on the production of fogs with ozone, since some observers considered that solid nuclei must have been present when condensation occurred, and others that all solid particles had been removed by passage of the gas through washbottles, etc.

On the technical scale absorption of fumes is very important, especially in the case of acid works, smelters and gas works (10, 11, 6, 4).

It has been pointed out by Knietsch (9) that the case of absorption of sulphur trioxide is determined to some extent by the rate of cooling of the gases. The clouds formed by rapid cooling were much more difficult to absorb than those formed by slower cooling. In the former case the particles will be smaller and will consequently have a much greater specific surface on which adsorption of air may occur. This layer of adsorbed gas is probably the chief factor in determining the case of absorption of the particles. And in the contact process the particles are dry and very difficult to wet. Reese (10) has illustrated this by passing air carrying sulphur trioxide through two washbottles in series, the first containing dilute acid and the second concentrated; a fog not absorbed by the strong acid is formed. If the order is reversed no fog is formed; showing that the strong acid apparently destroys the adsorbed gas film and thus permits wetting of the particles to take place. Remy (11b) has found that the percentage absorption of (contact) sulphur trioxide in sulphuric acid of different concentrations runs parallel to the boiling point, a maximum being obtained in each at about 98 per cent. acid. It has also been found that moist (i.e. containing sulphuric acid) fogs are absorbed better in water than the dry fogs, probably because the protecting layer of gas in the latter case renders wetting difficult. For example Remy (11c) found that with a fog from which there was absorbed 46 per cent. when moist, only 7 per cent. was absorbed when dry. But concentrated salt solutions absorbed the dry better than the moist (11a). When these fogs are tested by other methods e.g. filtration, the results are not so simple. When passed through a filter paper 71 per cent. of the moist fog was retained but only 36 per cent. of the dry When, however, the gas was passed through gas mask carbon 22 per cent. of the moist fog and 82 per cent. of the dry fog were removed. In these cases the explanation presumably is that the particles in the moist fog are larger than those in the dry, and consequently their rates of diffusion are much lower.

Remy has also determined the amount of absorption taking place when these fumes are passed through alkaline solutions, both

in the presence of an indifferent gas (air) and of an acid gas (carbon dioxide). He finds that in 20 per cent. potash the absorption is much lower than in pure water whether the fog is dry or moist. When a 1 : 1 mixture of air and carbon dioxide is substituted for air alone a considerable increase in absorption occurs; but in this case it is necessary to remember that as the bubbles of gas bearing the particles pass through the liquid they will be continuously decreasing in size, and hence better contact with the solution will be obtained. The following table gives some of Remy's figures (11c):—

In the last two experiments the fog was passed through five washbottles instead of one as in the previous cases.

Washing the air from concentration plants with dilute soda ash solution is said (2) to remove the particles of acid quite satisfactorily, although this would not be expected from Remy's work given in Table 1. (See also 8).

Ammonium chloride fogs have been examined by similar-methods and, in general, similar results have been obtained, except that the dry fogs seem to be absorbed more easily than the moist, probably because of their greater mobility. Caustic potash again causes a decrease in absorption, the dry fog being absorbed more easily than the moist (11a, b, c).

In 1898 Wilson (18) carried out experiments with clouds consisting of fine particles produced by an air blast impinging on the surface of different solutions. Among the solutions which he used were H²SO⁴, KOH, NaCl, and the non-electrolytes sugar and glycerine. Not only these solutions but all in which a non-volatile compound was dissolved behaved in the manner described below. After formation the spray was passed through concentrated sulphuric acid and then through water; in the space over the acid no particles were visible, but in the space over the water the cloud reappeared, except with a distilled water spray. The explanation of this behaviour is that the water is removed from the particles by the strong acid leaving very small solid nuclei which are too small to give a visible cloud, but which on again coming into contact with water can by reason of their hygroscopic nature cause condensation to occur and hence the reappearance of the cloud. Wilson found that hydrochloric acid below 5 per cent. and acetic acid did not give the above results. The density of the clouds in the flask containing the water was apparently proportional to the molecular concentration of the dissolved material and was equal for equal con-

centrations of the different substances, although a considerable change in the concentration had to be made to produce an appreciable change in the density of the fog. After electrification this re-appearance of the cloud could not be obtained, although the electrification was not removed. This would show that the charge remains on the nucleus after evaporation of the moisture, as is to be expected from the results of Gallrotti (quoted in 5), and owing to the slow diffusion of the nucleus the charge is not lost.

The above experiments show why ozone fogs behave as did Wilson's sprays after drying and subsequent contact with water vapour (28).

Now although the above experiments have shown that absorption by bubbling is very incomplete, no information is given as to what effect passage through liquids has on the composition of the particles. The change in particle size has been investigated (11b) for the above mentioned fogs and is considered in a later section of this summary. Some experiments by the present author have shown that even by passing a fog through one washbottle a very considerable change in the concentration of the dissolved material may be effected. A 50 per cent. (by vol.) alcoholic soda solution (1.24 N) was employed for generating a fog with hydrochloric acid. When the resulting fog was passed through a washbottle, according to the rate of passage of the gas, the absorption of the fog varied between 72 and 44 per cent., the higher figure being obtained with a narrower jet and a slightly lower rate than the lower. The most noticeable effect was, however, the reduction in the acid concentration of the fog particles. This was reduced, according to the conditions of the experiments, from 5.1 and 4.5 per cent. to 2.0 and 1.6 per cent. respectively. Since only about half of the particles were retained by the solution, the above reduction of acid must have been caused by condensation of vapour on to the particles with consequent increase in size. No definite experiments are however available to show whether it is so, but Remy's experiments would indicate this; nor were analyses made of these fogs to see if this condensation was preferential i.e., whether alcohol or water was the more easily condensed. It is hoped that this question will be examined more fully at a later date.

Allied to this question of absorption is that of methods of filtration of clouds and smokes. Owing to the very small particles filtration presents many difficulties. The best methods for the removal of very small particles appear to be thermal filtration (3) and electrical precipitation. Tolman has found that even the small particles of tobacco smoke are readily removed. In a temperature gradient the particles always move towards the cold end (1); consequently in Tolman's experiments the smoke particles were deposited on the cold tube of the filter. Electrical precipitation is not considered further in this paper.

In chemical warfare the removal of the particles of smoke and gas screens from the air before being breathed is very important. It has been found that there is a minimum efficiency for a filter at a certain size of particle, i.e., the filtration is more perfect with particles smaller or larger than this critical size (7).

stored over mercury: the pure dry phosgene did not attack the mercury. The head of a gas burette was fitted with a three-way tap, one branch of which was connected with the system of flasks. This outlet tube had a further three-way tap by means of which the gas could be passed either through the charcoal tower and through a mixing tube or direct to this tube and thence to the absorbing solution. Dry air was drawn through the other outlet tube of the burette and mixed with a definite quantity of phosgene vapour. This mixed gas contained one per cent. by volume of phosgene.

Phosphorus oxychloride vapour was freed from chloride and hydrogen chloride by passage over zinc dust and then dried by calcium chloride. Air carrying the vapour could pass either through the charcoal or direct to the solution.

Hydrogen chloride vapour was obtained by passing dry CO² free air through the concentrated acid followed by drying of the mixed gas with calcium chloride: frequent renewal of the acid was necessary owing to the rapid reduction in the partial pressure with slight decrease in concentration in the liquid.

The charcoal chiefly used was an active one in thin flakes, and was used in both its ordinary state of equilibrium with the atmosphere and its well-dried state. It gave, on analysis, moisture (loss at 160°C.) 13.54 per cent., ash 3.39 per cent. Before being used for the experiments the charcoal was saturated with the pure dry chloride vapour, dry air being then drawn through for a time to remove excess vapour.

The soda solutions were made up from sodium hydroxide “ex sodium” by dissolving a weighed amount in water and precipitating the carbonate present by a slight excess of baryta. After settling the clear liquid was decanted off. The alkaline solutions of the organic compounds used were made up by volume; for the more dilute solutions dilution was carried out with aqueous soda of the same concentration. In all cases the soda was approximately 1.25N.

The jets had their tips ground square and were chosen from a number for the regularity of their diameters and wall thicknesses. The internal radii were measured with a travelling microscope.

The term “depth of liquid” used later refers to the distance between the tip of the jet and the surface of the solution.

In carrying out an experiment the mixed gas was led, either after passing through a column of charcoal or direct through one (or two) washbottles containing the alkali solutions, any fog that was formed being filtered out by “woolly asbestos” placed in a stoppered U-tube which was weighed before and after an experiment. This arrangement was found to form a very efficient filter (for the practice and a theory of filters for fogs see Tolman et al. (J. Am. Chem. Soc., 1919, 41 299). After a given volume of gas had been drawn through, the apparatus was flushed out with air. Blockage of the tips of the jets by sodium chloride sometimes occurred, particularly when the stronger alcoholic solutions were used.

In a number of cases the chloride in the absorption flasks was estimated whence the original partial pressure of the hydrogen chloride (when this was used) could be calculated. The asbestos

particles remains almost constant. Approximately 3.2 per cent. of the total chloride carried by the gas stream appeared as fog. All weights of fog are given in mg. per litre of air.

When the fog formed in the first flask was carried through a second before the gas was filtered, not only did a reduction in density occur, but the concentration of acid in the particles was much reduced also. In the column headed “original fog” in Table 2 are given the quantities (some of which were obtained by extrapolation from other data) which are formed in the first flask under the conditions specified, and under “per centage absorption” the per centage of this fog which is absorbed by the second washbottle.

The data given show how much more effective is the narrower jet in reducing both the density of the fog and the acid concentration in the particles (compare Table 1).

The symbols 1 and 2 refer to the first and second washbottles respectively.

The whole of the reduction in density can not be attributed to the influence of the jet as an unknown amount settles out on the sides of the flasks and on the walls of the tubes through which the air passes.

With aqueous solutions experiments in which the depth of liquid was varied have been carried out. The flasks in which the acid was absorbed were of such dimensions that using the same volume of soda solution in each case (in order to keep the concentration gradient constant) the different “depths” of liquid could be obtained. In some cases only a small quantity of fog was formed, so that the analyses are perhaps not very accurate, this being the reason for the high per centages of chloride given for some of the fogs.

From the results given in Tables 3 and 4 it will be seen that with a given jet in every case the quantity of chloride escaping absorption and of fog formed increased as the rate of passage of the gas increased.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

The jets did not behave alike, however; for at 3.75 cm. the quantity of fog and of chloride at a given rate increased as the radius of the jet decreased, while at 7.4 cm. the opposite was the case. At an intermediate depth, 5.25 cm. the quantity of fog formed at a given rate with the two narrowest jets was almost independent of the radius.

When the data given in the above tables were graphed, plotting mg. of fog against rate of passage of gas, it was found that at 3.75 cm. all the lines were curved, but that at 7.4 cm. there was a transition according to the radius of the jet from practically straight lines to lines with a distinct curvature. The lines for 5.25 cm. had a very decided curvature, but were intermediate in shape between those corresponding to the other depths.

It is also seen that the concentration of acid in the particles remains very steady, especially for the fogs formed at the greater depth of liquid; in this latter case the concentration is slightly lower than that for the shallower liquid.

When the aqueous solutions are used only 0.8 to 1.5 per cent. of the acid vapour originally present in the gas stream escapes absorption, but the concentration of acid in the particles is considerably higher than with the alcoholic solution.

In Part 3 the partial pressure of the water vapour in the air in contact with the fog particles will be considered.

All the experiments described above had been carried out using an “inverted jet,” that is, used in the same manner as the inlet tube of an ordinary washbottle. It was desired to calculate the time of contact of the bubbles with the solution, but owing to the manner in

stant pressure apparatus to the aspirator. The levels of the liquid in the manometer were kept the same during an experiment. In cases of difficulty in getting this setting, the constant level apparatus was set in operation, water entering until atmospheric pressure was restored; the volume of water which entered was deducted from the volume of water run out of the aspirator. The rate of outflow was checked by a stop-watch in all coses. A filter-pump was arranged so that it could be used for suction from the aspirator, and when this was not desirable the air to be pumped into the reservoir entered through an inlet on the pipe-line. The different absorption vessels were of such sizes that by always using the same volume of solution, “depths” of liquid of 7.5, 5.5, 3.7 cm. respectively were obtained.

To carry out an experiment the three-way taps are opened, so that air only passes through the jets. It is very important to have air passing through while the solution is being poured into the absorption vessel, else liquid enters the jet. After the liquid has all been poured in, the taps are turned off until a bubble of air remains on the tip of the jet. The acid gas is then allowed to pass in at a rate regulated by a tap on the aspirator. The thermometer in the inlet tube gives the temperature of the gas mixture (practically air temperature) and a thermometer in the solution gives the temperatures at the beginning and end of an experiment.

When the “erect” jets were first used difficulties were encountered, owing to the working-back of the liquid into the barrel of the jet and thence to the inlet tap, but by coating the inside surfaces of the jets with paraffin wax this was obviated. The tips of the jets were cleaned, so that a glass surface was exposed for the formation of the bubbles.

The alkali solutions, whether of sodium, potassium or barium hydroxides, were free from carbonate.

In all cases the partial pressure of the hydrogen chloride is taken as 130 mm. Although no difference in the amounts of fog formed under similar conditions from moist or dry gas were found, in all cases the gas was dried by calcium chloride.

The fog was analysed by the method previously given. The “partial pressure” of the hydrogen chloride given in the tables has been calculated by taking one mg. of hydrogen chloride as giving a pressure of 0.5014 mm. Hg. at 20° C., or 0.5098 mm. at 25° C. From the increases in weight of the calcium chloride tubes, assuming that water vapour obeys the gas laws, the partial pressures given in the tables were calculated.

In all cases “per cent. HCl” refers to the percentage of acid by weight in the fog particles. All weights of acid and of fog are given in mg. per litre of air used. The amount of free acid in the particles has been determined in each case as a check on the silver titration, but is not given in the tables.

Owing to the variation of the amount of fog formed with change in concentration of the alkali, it was necessary to use the smallest possible volume of air in the experiments in which the “rate” was altered. Two litres has been adopted as standard, as with this volume the concentration has not been too greatly reduced and a weight of fog sufficient for analysis has been obtained.

(a) The necessary presence of alkali: It has been repeatedly observed that in neutral or acid solutions of salts, dyes and colloids, and in distilled water no fogs are formed under the conditions of the present experiments. This is clearly shown also in the data obtained when the effect of alkali concentration on fog formation was being studied—immediately the solution became acid, fog formation ceased.

A number of salts of weak acids were examined also. Solutions of the following (in normal solution generally), sodium acetate, bicarbonate, disodium phosphate, potassium dihydrogen phosphate, and sodium biborate gave no fogs. Thin fogs were obtained with sodium carbonate (IN), and good ones with trisodium phosphate (IN). These results show that a certain hydroxyl-ion concentration is required before fog formation begins.

(b)Alteration in radius of the jet and in the depth of the liquid: In practically every case an increase in the radius of the jet causes a greater production of fog. It would appear that at about 5.5 cm. depth the production of fog is at its maximum, for here the two widest jets give considerably more fog, especially at the higher rates, than at either 3.7 cm. or 7.5 cm. With the narrowest jet, there is comparatively little difference in the amount of fog produced at each depth of liquid. With the other jets there is a tendency for less fog to be produced at 3.7 cm. than at 5.5 cm., or 7.5 cm., especially at the higher rates, although the variation is only a few milligrammes per litre of air.

As is noted later, the concentration of acid in the particles is greatest at 3.7 cm., so that the weight of hydrogen chloride escaping absorption decreases as the depth of liquid increases, except for the two widest jets which show maxima at 5.5 cm. On plotting the data obtained, it is found that as the depth of liquid increases and the radius of the jet decreases, the amount of fog formed, and especially the weight of acid vapour escaping absorption tends to vary linearly with the rate of passage of the gas.

(c)Alteration in rate of passage of the gas: In all cases any increase in the rate of passage of the gas causes an increase in the amount of fog formed. With the simple aqueous solutions, in general, this increase is not a linear function of the rate. There is evidence that below a certain critical rate the fog production falls off very rapidly, but owing to experimental difficulties, when only small quantities of fog are obtained, it is not certain whether this critical rate exists for solutions containing no colloid or dyestuff.

It may be stated here that the fog formation is not due merely to the comparatively rapid rates at which the gas is bubbled through the solutions, for with gelatine, etc., solutions, a thick fog is formed when only one bubble every one or two seconds passes up through the liquid.

As an example of the results obtained when the radius of the jet and the rate were altered, in Table 2 are given data obtained when a 1.26 N sodium hydroxide solution was used.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

(d) Alteration in concentration of the alkali: A large number of experiments have been made in which the same alkali solution has been used until completely neutralized. As shown by the representative data in Table 3, the weight of fog formed has been determined after the passage of each litre of air through the solution. In all cases the fog production falls off as the alkali concentration decreases. As soon as the solution becomes acid, no more fog is formed.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Solutions of potassium hydroxide and barium hydroxide give similar results, although at corresponding concentrations less fog is obtained than with sodium hydroxide.

If the logarithms of the mean concentrations of alkali during the passage of each litre of air and of the corresponding weight of fog be taken, a straight line is obtained when they are plotted. Moreover under fixed conditions of depth and original concentration of alkali the lines corresponding to the different jets are parallel.

From the data obtained in these experiments Table 4 is compiled, showing the amount of fog formed in the first litre of air passed through sodium hydroxide solutions of the mean concentrations given at 20° C.

If these figures are plotted in the ordinary manner it will be seen that in dilute solutions small additions of alkali exert very large effects; in the more concentrated, the amount of fog obtained is almost proportional to the alkali concentration.

The experiments made with the sodium carbonate and trisodium phosphate solutions may now be considered. The data obtained are given in Table 5, the rate being 225 c.c./min., and the depth of liquid 7.5 cm.; at 22° C

There is a rapid fall in the amount of fog formed as the solutions are neutralized. Buffer action may possibly account for the constancy of some of the carbonate figures; the phosphate solution reaches the disodium stage too quickly for the above to be observed, but it doubtless exists in this case also.

Sodium carbonate is not so greatly hydrolyzed as sodium phosphate, and consequently the hydroxylion concentration will be somewhat smaller in the carbonate than in the phosphate solution. This difference is plainly shown by the differences in the amount of fog obtained from the first litre of gas. Now, from the data given in Table 4, the concentrations of free sodium hydroxide which would give the same weight of fog under the same conditions are 0.06 and 0.45 normal respectively; these are of the same order of magnitude as would be expected from the degrees of hydrolysis.

It has not been possible to fix a definite value for the hydroxylion concentration below which no fog appears, but it must be greater than 10^-5 normal, since the other phosphates and sodium bicarbonate give no fog.

(e) Effect of the above factors on the concentration of acid in the fog particles: The acid concentration remains remarkably constant, varying only from 8 to 11 per cent., although the radius of the jet, the rate of passage of the gas, and the depth of the liquid have all been varied. At the two depths, 3.7 cm. and 5.5 cm., for a given rate the concentration of acid is the same; at 7.5 cm. for the same conditions the acid content is about one per cent. lower. At 3.7 cm. the acid content is almost independent of the radius of the jet; in the other cases there is a tendency for the acid content to fall as the radius of the jet is decreased. In all cases there is a slight tendency for the concentration to fall as the rate is reduced.

(f) Water vapour partial pressures: In all cases, after making due allowances for differences in temperature, the pressure of the water vapour in the filtered air remains fairly constant at about 95 per cent. of the saturation value for the corresponding air temperature, or 80-90 per cent. of that for the mean temperature of the solution during an experiment. When large quantities of fog are obtained the relations are not so simple, as in some cases values of more than 100 per cent. saturation are obtained. This point is considered later, when dealing with the effect of adding colloids, dyes, etc., to the alkali solutions. With small quantities of fog the vapour pressure is independent of the nature of the solution, the radius of the jet, the depth of the liquid, and the rate of passage of the gas. This presumably shows that the particles attain some sort of equilibrium with the vapour with which they are in contact. This point will be considered more fully when dealing with the sizes of the particles in the fogs.

2. During some of the experiments it was noticed that traces of certain organic compounds exerted a marked positive effect in fog production. For example, in an experiment with a 0.313N sodium hydroxide solution 10 drops of a 0.02 per cent. aqueous solution of methyl red were added, in order to see at what stage all the alkali was neutralized, but much more fog than usual was obtained, namely, 36.7 mg., instead of 22.2 mg. Hence it appeared to be of interest

to obtain strictly quantitative data on the effect of small quantities of colloids or capillary active substances on fog formation.

The following have been used as “capillary active substances”: iso-amyl alcohol, sucrose, gelatine, gum arabic, safranine, methylene blue (hydrochloride), congo red, orange G, glycerine, saponin, kaolin, aquadag, and clay.

Solutions or suspensions of these substances in sodium hydroxide were made up, and the amount and the analysis of the fog formed were determined for the different experimental conditions. The analyses given in experiments on the progressive neutralization of the alkali, except where otherwise stated were carried out on the total amount of fog formed, and are therefore average values.

Finally, most of the substances were examined with the conditions of experiment fixed, except for variations in the concentration of the capillary active substance.

Gelatine and Gum Arabic: The solutions were made up by allowing a weighed quantity of the solid to disperse in the alkali solution. Both the gelatine and gum arabic gave no fogs in absence of alkali.

In the case of gelatine solutions of different concentrations were made up, one of which was 0.5N with respect to sodium chloride. It was found that the amount of fog was not very dependent on the concentration of gelatine in the stronger solutions, but even traces of gelatine have a very great effect relative to the amount of fog obtained with a simple aqueous solution. Some of the data are given in Table 6.

Addition of sodium chloride to the solution caused a slight increase in the amount of fog formed.

Small alterations in the concentration of the alkali did not have a great effect on the formation of the fog.

When the rate of passage of the gas was varied, at about 110 c.c./min., the fog production decreased rapidly. This is shown by the data in Table 7. This peculiarity is shown by other solutions also.

A number of experiments were carried out at different rates and with different jets and solutions to find out how the fog production varied during the neutralization of the alkali. An example is given in Table 8. The solution was 0.591N sodium hydroxide containing 0.25 per cent. of gum arabic.

Iso-amyl alcohol, saponin, glycerine and sucrose: The amyl alcohol had a very definite action in increasing the amount of fog. In a 0.594N sodium hydroxide solution containing 1.0 per cent. of alcohol, at a depth of 7.5 cm., the gas being passed at 225 c.c./min., using the jets of radius 1.32 and 0.82 mm., from the first litre of air 79.3 and 83.7 mg. of fog respectively were obtained. Even in 0.01 per cent. solution its effect was quite definite.

Saponin was used in 0.01 per cent. solution. Some difficulty was experienced due to frothing. This solution gave three times as much fog as a simple aqueous solution under similar conditions. The acid content of the particles was somewhat less than in other cases, namely, 6.5-7.0 per cent.

Glycerine and sucrose did not have any effect on the amount of fog formed.

Safranine and methylene blue: Safranine was used in 0.61N sodium hydroxide. It has been found to be of the same order of activity as gelatine. For experimental conditions such that a pure aqueous alkaline solution gave 31.9 mg. of fog, the 0.02, 0.008, 0.0013 per cent. solutions gave 117.9, 83.2, 37.3 mg. respectively. Analysis of the fogs gave a percentage of 8.89 of hydrogen chloride.

Methylene blue was extremely active, so that a dilute solution containing 0.0013 per cent. was used as a stock solution for dilution in some of the experiments. The dye was absorbed very strongly

by all glassware, etc. This property seemed to run parallel with activity in fog formation since safranine and methylene blue were much more strongly absorbed than congo red or orange G.

The same phenomenon was noticed here as with the gelatine solutions, namely, that in the higher concentrations, the chiloride in the fog particles was somewhat greater than in the more dilute; for example, the fogs from the strongest solutions gave on analysis 10.2 per cent., and the fogs from the more dilute 8.9 per cent. For quantitative results see Table 21, when it will be seen down to what extreme dilutions the effect of methylene blue can be detected.

Congo red and orange G: Neither of these substances was very active, although the effect of their presence could be detected easily in a 0.0067 per cent. solution. For conditions such that 31.9 mg. of fog were formed from the simple solution, congo red and orange G each in 0.1 per cent. solution gave 71.2 and 90.2 mg. respectively. The percentages of acid were 8.11 and 8.34 per cent. respectively.

Suspensions: Colloidal solutions may be considered as very fine suspensions, so that it is of interest to see what effect a coarse suspension would have on the absorption of the acid vapour. With suspensions of kaolin, graphite (aquadag), and clay, dense fogs were obtained in each case.

Two different concentrations of acid vapour have been used, the partial pressures being 36.7 and 139.2 mm. Experiment has shown that not only is the amount of fog formed, but also is the concentration of acid in the particles very much dependent on the original partial pressure. For example, in a solution containing 0.039 per cent. of aquadag the weights of fog were 88.7 and 129.6 mg. respectively, the corresponding concentrations of acid in the particles being 7.22 and 8.55 per cent. In Table 9 are given some of the results obtained with kaolin. For aquadag, see Table 21.

A sample of carefully washed and fractionated Buxton clay was used. The particles were very small, taking some days to settle a few centimetres. This suspension was very active, as is shown by the following figures quoted from the full experimental results.

With a 0.124 per cent. suspension 208.0 mg. of fog obtained, and even with a 0.006 per cent. suspension (which appeared very slightly cloudy to the naked eye) 106.5 mg. of fog were obtained. The mean concentration of acid in the particles was 9.56 per cent.

All the experiments with the suspensions have been carried out under the same conditions namely, with a jet of radius 1.32 mm., and the gas passing at 208 c.c./min., except for the first set of experiments with kaolin, when the rate was 204 c.c./min.

The activity of aquadag approaches, and of Buxton clay exceeds, that of gelatine or safranine, while the kaolin has the same order of activity as congo red.

3. Hydrobromic acid has been found to be much more active than hydrochloric acid, this being parallel to its greater “fuming” properties. Neither hydriodic nor nitric acid has yet been tested, but each would doubtless give fogs.

As only a small quantity of hydrobromic acid was available, it was not possible to do many experiments, and, in addition, the partial pressure of the bromide vapour was smaller in each successive experiment; but the partial pressure was estimated in each case.

In all the following experiments the jet of radius 1.32 mm. was used at a depth of 7.5 cm., the gas being passed at 208 c.c./min.

Distilled water gave no fog with a gas in which the partial pressure of the bromide was 79.7 mm. The water vapour partial pressure was 19.4 mm., the air temperature being 21.7° C. This vapour pressure is the same as that found later when fog formation was taking place.

Two different sets of experiments have been carried out using the same concentration of soda (1.26N), but with the fresh solution and the same after the bromide partial pressure had fallen to about 70 per cent. of its original value. It will be noticed that the fog obtained at corresponding intervals in the second set is about two-thirds of that in the first set, so that the production of fog has fallen off almost in proportion to the reduction in the partial pressure of the bromide. The data of these experiments are given in Table 10, the figures in the column headed “mean” being obtained by analysis of the whole of the fog collected, and of the alkali solution at the end of the set of experiments.

When the same gelatine and gum arabic solutions were used with hydrogen bromide as had been employed with hydrogen chloride, very dense fogs were formed, the bubbles as they passed through the liquid being filled with fog. The concentration of acid in the particles remains practically constant, whether the solution contains any colloid or not.

In Table 11 the data obtained in experiments on the production of fog with progressive neutralization of the alkali are given. In each solution the sodium hydroxide was 0.59N. The “mean” values were obtained as in Table 10.

The percentages of bromide in the particles are very large; but when these are converted to normalities the concentration is nearly the same as in the case of the chloride, being, in fact, slightly higher. This would appear to show that the particles in each case are of the same size. How far this is true will be seen later when the sizes of the particles are considered.

In the gas, after removal of the particles, the water vapour pressure is constant, and has the same value under corresponding conditions as with hydrogen chloride, although with the gelatine and gum arabic solutions there is a tendency for it to be somewhat greater than in the previous experiments.

If the quantity of fog obtained be plotted as has been done for hydrogen chloride, the same type of curve is obtained. Corresponding to the greater production of fog in these experiments and to the slower falling off as the alkali is neutralized, the lines are not so steep as for hydrogen chloride. It is interesting to note that the lines for gum arabic and gelatine are again practically parallel.

The vapour pressures of the bromide in all these experiments have been considerably lower than with the chloride, yet the amount of fog formed under the same conditions is greater. This is probably due to the lower rate of diffusion of the bromide in the gaseous phase.

4. Although alcohol might have been included among the “capillary active” substances, it does not appear to be particularly effective. For example, a 7.6 per cent. solution gives an increase of only about 25 per cent. over the plain aqueous solution of the same alkali con-

centration. The strong alcoholic solutions give dense fogs, which begin to form as the bubbles pass up through the liquid. In this case the presence of a volatile substance lends interest to the experiments.

The solutions used contained 7.6 and 33.0 per cent. by weight of ethyl alcohol, and were 1.223N and 1.238N respectively, with respect to sodium hydroxide.

Alcohol was estimated by the method of Benedict and Norris (J. Am. Chem. Soc. 1898, 20, 293), in which a solution of potassium dichromate in sulphuric acid is used. The procedure finally adopted was as follows. A known volume of the very dilute alcohol solution (0.1 per cent. or less) is run from a burette into a small flask; three times this volume of concentrated sulphuric acid are then added slowly with cooling under a good stream of water. An excess of dichromate solution is slowly added from a burette, after which the flask is heated over a small flame, taking fifteen minutes to reach 98° C., the temperature then being held at this point for five minutes. The flask is then cooled and added to about 300 c.c. of cold water in a large conical flask. After again cooling, the excess of dichromate is destroyed by adding an excess of standard ferrous ammonium sulphate, this excess being in turn estimated with standard permanganate.

The dichromate was standardized by means of a 0.100 per cent. solution of alcohol. This solution was of approximately the same strength as the solutions obtained after extracting the asbestos filters with water.

Since the presence of chloride causes high results, the following procedure was adopted. The asbestos was removed and extracted with distilled water, the volume being kept as small as possible (not more than 100 c.c. with the smaller quantities of fog), the washings being run into a graduated flask. An excess of 0.03185N silver sulphate solution was then added, and, after being well shaken the liquid was made up to the mark. After standing for a time, the solution was filtered, and an aliquot taken for analysis. Another portion of the solution was then titrated with ammonium thiocyanate to determine the excess of silver. From the gross amount of alcohol found on analysis a deduction was made for alcohol absorbed by the asbestos, this having been determined in blank experiments.

An attempt has been made to estimate separately the alcohol and the water vapour partial pressures in the gas after the fog particles had been filtered out. The most satisfactory method for the conditions under which it was required to work, appeared to be that of Thomas (J. Soc. Chem. Ind., 1922, 41, 33T). In this method freshly broken calcium carbide, which must be as free as possible from calcium oxide, is used to remove the water vapour, the alcohol passing on to be obsorbed in concentrated sulphuric acid. The factor for converting the change in weight of the carbide to water was found by blank experiments to be 2.91. Blank experiments showed that no acetylene was absorbed by the acid.

In all the experiments where fogs were formed, the partial pressures were estimated, whence the ratio water/alcohol was calculated, but the results in some cases were somewhat irregular. The ratios

water/alcohol found in the vapour correspond satisfactorily with those given in the standard tables for mixtures similar to those employed in these experiments.

The details of a typical experiment are given:—

The fog was washed out into a 250 c.c. flask containing 15 c.c. of the silver sulphate solution. After filtering, 100 c.c. of solution required 2.15 c.c. of ammonium thiocyanate solution (equal to the sulphate in concentration) for precipitation of the excess silver; whence the chloride present used 9.63 c.c. of silver solution. This represents 11.18 mg. of hydrogen chloride so that the particles contained 4.94 per cent. by weight of chloride.

For the alcohol estimation two portions of 25 c.c. each were taken, to which were added as described above, 25 c.c. of the chromic solution. To destroy the excess chromic, 10 c.c. of a N/10 ferrous solution were used, for the titration of the excess iron 3.27 c.c. of N/10 permanganate were required; hence the alcohol was oxidised by the equivalent of 10.95 c.c. of the ferrous solution, so that in 250 c.c. of solution there were 109.1 mg. of alcohol (since 1 c.c. of the ferrous solution was equivalent to 0.996 mg. alcohol). The correction for adsorption was 10.3 mg., so that finally the fog contained 98.8 mg. alcohol giving a percentage by weight of 43.6.

Series of experiments similar to those already reported for the other alkaline solutions have been carried out. Some of the results are given in Table 12. It will be noticed that with the weaker solution there is not a great deal of difference in the amounts of fog formed at corresponding rates with the two jets employed; but with the stronger solution the narrower jet gives the greater amount of fog. This has been noticed for the still stronger solution used in Part 2 of these experiments. Also, in the weaker solution, the concentration of acid in the particles approaches much closer to that found with the ordinary aqueous solutions than does that with the stronger solution. In the latter case, the figures are in agreement with those found in Part 2. Owing to the difficulty of preventing liquid from creeping back into the jet, no results could be obtained with the widest jets.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

A peculiarity that was shown by varying the rate of passage of the gas was that although at higher rates the amount of fog was almost proportional to the rate, at the lower rates the amount of fog produced fell off much more rapidly than this straight line law required. The data of the above tables show this.

The fogs produced at each stage of the neutralization of the stronger solution have been analyzed. As found with other solutions, the composition of the particles and of the vapour remains practically constant throughout. The data in Table 13 illustrate this. The composition of the particles and of the vapour remains constant when the rate of passage of the gas is varied as shown by Table 12.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

In connection with these experiments it will be noticed that the percentages of alcohol found in the particles are somewhat greater than those in the original solutions. In view of the work of Kablukow (Zeit. phys. Chem. 1903, 46, 399) it is probable that the composition of the vapour from the alkaline solutions is richer in alcohol than that given by a simple aqueous solution of the same alcohol content. If a correction is applied, assuming that sodium hydroxide will have the same effect on the vapour composition as has potassium iodide, then it is found that the percentage of alcohol to be expected in the liquid in equilibrium with the vapour given off is close to that actually found.

It is well-known that pyridine fumes strongly in the presence of hydrogen chloride, so that it is not surprising that dense fogs are formed when air carrying the acid gas is bubbled through a pyridine solution.

A M/2 solution of pyridinc was used. Experiments on the progressive neutralization of the solution showed that fog formation ceased only after the solution became distinctly acid. This is probably due to the hydrolysis of the pyridine salt allowing vapour to be given off to form a fog until a sufficient acid concentration is built up to repress the hydrolysis. The bubbles rising through the liquid were filled with fog, so that in this case, as with some of the other solutions already mentioned, at least part of the non-absorption of the acid would be due to the slow diffusion of the hydrogen chloride condensates and in accord with the known difficulty of removing suspended matter from a gas by bubbling. The composition of the particles remained constant during the neutralisation of the base. No determinations of the pyridine content of the particles have been made. Some of the data obtained are given in Table 14.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

This pyridine solution was also used in experiments in which the rate of passage of the gas was varied. With both the jets employed it was found that at the higher rates, say, above 110 c.c./min., the amount of fog obtained varied linearly with the rate, but below this rate the formation of fog rapidly decreased. (See Table 14 a). In this case also the narrower jet was the better fog producer. From Table 14(a) it will be seen that the acid concentration in the particles falls as the rate decreases, although the water vapour partial pressures remain constant. In some cases the pressure is greater than corresponds to saturation at the given temperatures, but this is probably due to the temperature of the gas being higher than that of the solution, owing to the latent heat of condensation of the water in the large quantities of fog which are obtained.

5. (a) Sizes of the Particles: The chamber in which the fog was allowed to settle was constructed as follows: An inner cylinder of thin, hard glass was surrounded by a thick outer one, the space between the two being sealed at each end. Wide bore taps were used on the inlet and outlet tubes, in order to eliminate as much as possible contact of the fog with any surfaces with consequent coagulation of the particles. Two narrow slits the length of the cylinder were made in the blackened outside wall; these slits were at right angles, one for the entry of light filtered through a water cell, and the other for observation of the level of the fog column. The space above the fog appeared black, while the fog appeared white, the intensity of the dispersed light being dependent on the density of the fog. Along the last-mentioned slit was placed a calibrated glass scale by means of which the height of the surface of the fog could be read at any time. At each 2.5 cm. the times taken for the surface to fall from the zero mark were read on a stopwatch. The surface of the fog remained level during an experiment, except when the densest fogs were being observed.

The following data have been employed in the Stokes-Cunning-ham equation used for calculating the radii of the particles from their rates of fall: for aqueous, 7.6 and 33.0 per cent. alcoholic, pyridine, and hydrobromic acid solutions, the radius is calculated from r = X √v where X has the values 8.92, 9.00, 9.23, 8.92, 8.40 × 10^-4 respectively, the densities of the respective particles having been taken as 1.04, 1.02, 0.97, 1.04, 1.17.

Experiment showed that the radius was the same at all stages of the neutralization of the solution. But the particles increased somewhat rapidly in size during the time of settling, as the rates of fall were greater near the bottom of the scale than at the top. With the dense fogs the particles appear to be smaller than in the other cases. The radii given in the tables which follow have in such cases been determined from the rate of fall of a separate layer which sometimes appeared, or of the original surface at a later period of the settling process. Table 15 gives the radii of the particles in a number of dense fogs. In the first part of the table the solutions were 0.62N with respect to sodium hydroxide.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

An examination has also been made of the variations of the radii when the colloid, etc., the radius of the jet, and the rate were altered. In general the particles were larger as the concentration of colloid decreased, as the radius of the jet decreased (rate constant), and as the rate decreased (radius of jet constant). When the colloidal solutions or suspensions were used, the radii were distinctly less than with the ordinary aqueous alkali solutions. This may be seen from Table 16 where the data given are selected as being representative of the many obtained; in all the solutions the alkali was 0.62N sodium hydroxide.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

With the alcoholic solutions the particles given by the stronger solution were somewhat the smaller. In both cases the particles grow rapidly with very considerable thinning-out of the fog. This rapid growth has been observed previously by Barus (Carnegie Inst., No. 62, 1907, p. 113). In Table 17, “original” and “final” refer to the radii calculated for the beginning and the end of the readings.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Hydrobromic acid gave similar results to hydrochloric acid as is shown by the data of Table 18. In all cases the particles are smaller than those obtained under similar conditions with the chloride.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

From the various data obtained, it appeared that the original size of the particles is possibly of the same order as for fogs from other reactions, namely, 5 × 10^-5 cm.: for example, pyridine has given particles of radius 5.0-7.2 × 10^-5 cm., methylene blue of 7 × 10^-5 cm., and Buxton clay of 10 × 10^-5 cm., These have been found to be very unstable, often rapidly increasing to a mean radius of 2 × 10^-4 cm. Whether the comparatively large particles found for the ordinary aqueous solutions have grown from such smaller particles it is not possible to say.

It is to be noted that apparently an equilibrium state is reached by the particles, since the vapour pressures and concentration of acid in the droplets remain constant and independent of the condi-

the same as that found for the simple aqueous solutions. When the concentration of active material varies the relations are not so simple. In Table 19 are given some of the results of analysis of fog obtained when the concentration is varied.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

It is seen that the acid content of the particles is not independent of the material used.

From the various data obtained it can be stated that: (1) the concentration of acid in the particles tends to decrease as the concentration of “capillary active” decreases; (2) for a given concentration of “capillary active” substance the acid concentration is independent of the concentration of the soda solution; and (3) the pressure of the water vapour in contact with the particles is constant.

When the data obtained in the experiments with varying concentration of “capillary active” substance are plotted straight lines are formed corresponding to the general equation log F = m log C + K in which F is the weight of fog in mg. obtained from the first litre of air used, C is the concentration of added material, and m and K are constants. Except for methylene blue, where it is possible that a curve might be drawn through the points, and for congo red and orange G, where the points corresponding to the most dilute solutions lie off the lines, the above statement is strictly true. In Table 20 are given the values of m and K for the solutions examined, and in Table 21 are given the experimental values, together with the values calculated from graphs for three typical substances. It will be seen that in general the agreement is satisfactory.

Kaolin and aquadag (1): Partial Pressure of HCl 36.7 mm.

Kaolin and aquadag (2): Partial pressure of HCl 139.2 mm.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

The activity of the substances extends over a wide range. The only generalization, however, which it would appear possible to make

is that basic substances (e.g., safranine, methylene-blue) are much more active than others (e.g., gum arabic, congo red).

The great capillary activity of such substances as gelatine and methylene blue has has been noticed in other directions (Donnan Brit. Assn. Rep., 1923, Sect. B. 12), e.g., electromotive force (Sândera Rec. Trav. Chim., 1925, 44, 480), cataphoresis (van der Grinten J. Chim. Phys., 1926, 23, 209), water-fall experiments (McTaggart Phil. Mag. 1914, 27, 297). As the relation between the amount of material present and the effect it produces is of the same form as the “adsorption isotherm,” it has generally been concluded that adsorption at the interface is the cause of the phenomena.

It is rather remarkable, however, that such great effects should be obtained in the present experiments, since the surface tensions of the solutions differ very little from that of pure water, or from the simple alkali solutions. That surface tension is not the controlling factor, at any rate so far as fog formation is concerned, is shown by the fact that although amyl alcohol causes a great reduction in the surface tension, yet it is not very active as regards fog formation.

The use of pyridine is complicated by its volatility, and this is part of the reason why fog is formed in the bubbles as they pass through the liquid. But if pyridine vapour were passed through acid solutions, the result would be quite different.

This aspect has been examined qualitatively by using solutions of ammonia and of hydrochloric acid of varying concentrations. If a gas bearing the chloride vapour is passed through aqueous ammonia, fogs are formed in all cases, because even in very dilute solutions ammonia has an appreciable partial pressure. If, however, the vapour from even concentrated ammonia be passed through fairly strong hydrochloric acid solutions, no fog is formed until the partial pressure of the chloride rises above its infinitesimally small values in the more dilute solutions. In fact, it is only above about 18 per cent. acid that fogs begin to be formed; even at this stage the partial pressure of the acid vapour is only a fraction of a millimetre. As the concentration is still further increased, the fogs become more and more dense.

The mechanism of the absorption of the acid will now be considered in some detail.

If the data given in Table 1 are used to calculate the times of contact of the bubbles with the solution, it is found that these lie between 0.15 and 0.35 second according to the jet and the rate used. Now the widest jets give the largest bubbles, and these move more slowly through the solution. But calculation shows that allowing for change in volume, due to absorption of the acid vapour, for any pair of jets the ratios of the radii and times of contact are practically the same. This leads to the experimental result that the amounts of fog obtained in simple alkali solutions do not vary a great deal.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

According to the two-film theory of gas absorption the equation is dw/dt= AK (Pg-Pi)

where dw/dt is the rate of increase in weight of material dissolved,

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

A is the area of the bubble, K is a constant and Pg and Pi are the pressures in the gas film and in the liquid film respectively (Lewis and Whitman J.I.E.C., 1924 16, 1215). For hydrogen chloride and bromide Pi is negligible in dilute solutions, hence the equation reduces to dw/dt = AK Pg. Now dw = - X dp where X is a function of the molecular weight of the gas, and dp is the fall in pressure brought about by solution of the weight dw. Hence dp/dt = A K Pg/X which gives log_e+ Po/Pt = Akt/X where Po and Pt are the pressures at times o and t. Since in these experiments Po is constant, and A is constant for a given jet, the equation reduces to K—K't = log_e Pt where K and K′ are constants. Again since the concentration of acid in the particles is nearly constant the equation reduces finally to K″—K′t = log_e F where K” is a constant and F is the number of mg. of fog per litre of air. This requires a linear relationship between t and log_e F. Experiment agrees with the above theoretical requirement as is shown by the data in Table 22.

The lines drawn from data such as those in Table 22, for the different jets are practically parallel at 3.7 cm. and 5.5 cm., but at 7.5 cm. they become slightly steeper as the radius of the jet increases. Moreover, there is not a great difference between the slopes of the lines whatever the jet or the depth of liquid.

It may be useful to examine briefly the different factors which might be considered in being causative agents in the reduction of the rate of diffusion of the acid vapours into the solutions.

First, surface tension effects may be considered solutions of all alkalis and alkali metal salts raise the surface tension of water against air (Landolt Börnstein Tabellen 1923 Edn. 238 et seq.), and

when present with amyl alcohol, etc., cause much greater lowering than if the alcohol is there alone (Seith Zeith. phys. Chem. 1925, 117, 257). But, as mentioned above, amyl alcohol is not nearly so active in causing increased fog formation as solutions containing dyes and colloids, the surface tensions of which are not greatly different from the simple solutions (Freundlich and Neumann Koll. Zeit. 1908, 3, 80). Moreover, if alkáli is absent from such solutions no fog is formed. This shows that the decreased rate of absorption is bound up with the presence of alkali more than with surface tension effects.

It has been already seen that the gas film is the only one to be considered in the absorption of hydrogen chloride. There is the possibility that the rate of absorption is decreased by the presence of air in the mixture, since it has been found that the absorption of ammonia is very greatly affected in such a case (Kowalke et al. Chem and Met. Eng., 1925, 32, 506). Thus the rate of absorption from a mixture containing 40 per cent. of ammonia was only one-fiftieth of that for pure ammonia. But if such an effect were a predominant one, then fogs would be expected even in the absence of alkali.

The rate of absorption is also governed by the rate at which diffusion both in the gaseous and in the liquid phases occurs. It is necessary to consider only the liquid phase in detail. Now both hydrions and hydroxylions diffuse much more rapidly in salt solutions than in pure water (Lewis Syst. of Phys. Chem., 1920 2, 204) while gases diffuse more slowly, the reduction increasing with increase in concentration, of the salt (Barus Car. Inst. Pub., 186, 1913). Here again such an explanation of the fog formation breaks down as fogs would be expected, on the above grounds, in salt solutions.

All considerations of this problem lead to the conclusion that the alkali film is the cause of the decrease in absorption. A somewhat similar problem has been examined by Ledig (J.I.E.C., 1924, 16, 1231) in the absorption of carbon dioxide by alkaline solutions. Three stages in the absorption were noticed, first a very high initial rate which falls to a second fairly steady rate, and third a much slower rate probably governed by the now lowered partial pressure of the gas. In sodium carbonate the rate of absorption was much lower than in pure water. This was not wholly due to viscosity effects, as a cane sugar solution of equal viscosity absorbed the gas more rapidly than the carbonate. Again, a distinction was noted between potassium and sodium hydroxides, potash being the better absorbent—a fact which agrees with the present experiments that potash gives less fog under similar conditions than soda. Distilled water did not give the high initial rates obtained with alkaline solutions, pointing to the fact that the film around the bubble very rapidly becomes saturated with the gas.

Perhaps something of a similar nature occurs when hydrogen chloride is absorbed, although some explanation is required why no fog forms in the presence of neutral salts. Adsorption effects must be in operation on the surface of the bubble, but there will be a tendency for a fresh water surface to be kept there owing to the nega-

tive adsorption of the ions. Dyes and colloids will be positively adsorbed and will form a film through which diffusion must take place. If, as Alty and others have found (Proc. Roy. Soc. 1924, 106A, 315; 1926 110A 178; Phil. Mag., 1914, 27 297), the bubble is negatively charged then there will be a tendency for negative particles to be attracted into the positively charged side of the double layer, that is, there will be a layer of negatively charged particles in the liquid film round the bubble. Hence basic substances will be more active, independent of, but in conjunction with, adsorption. Since the heavy ions would diffuse more slowly than the hydroxyl ions, the rate of fall of pressure would be less than when they were absent. For some reason the presence of alkali in the solution intensifies very greatly the effect of this film of dye or colloid.

The fact that the vapour pressure remains constant and independent of the amount of fog requires some explanation. In the case of very active substances fog is easily visible in the bubbles as they pass up through the liquid. Once a fog particle is formed it will diffuse very slowly, and hence have very little opportunity of being absorbed by the alkali; but before this particle can form, since the gas mixture is originally dry, water vapour must diffuse rapidly into the bubble. Any such particle formation, or even the formation of molecular aggregates would tend to keep the partial pressure of the water vapour low and hence aid further diffusion of the vapour. The alkaline film with its adsorbed material thus appears to act as a membrane which is much more readily permeable to water vapour than to other gases or ions. That the surface of an alkaline solution may be very efficient in the evaporation of water, more so than a pure water surface, is known technically; also that an alkaline film can be impermeable to gases has been shown by Taylor (Fuel, 1926, 5, 195) who found that a layer of alkaline clay was impervious to the gases formed by decomposition of organic matter underneath it. As the whole system in this case was wet, it was not possible to observe if water vapour could pass readily or not. In this case, colloids would, of course, be present, and would form a layer of adsorbed material at the boundary between the soil and the vegetable matter.

This examination of the different factors shows that around the air bubbles an adsorption layer is formed in the presence of alkalis which does not readily permit the diffusion of the ions formed by the absorption of the acid vapour and thus causes a decrease in the rate of absorption to such an extent that the bubble reaches the surface with some of the acid vapour (in extreme cases up to 12 per cent. of the total) still unabsorbed. Capillary active substances in the solution increase the resistance to diffusion very considerably, although the diffusion of water vapour appears to be increased.

It has been previously pointed out that a common radius would be expected for the particles in the fogs, but the data in the tables given above show that this expectation is not wholly borne out by experiment.

The Stokes-Cunningham equation which, be it noted, gives only am average result for the radius as in general a large number of particles are observed at once, has been used for all types of fogs and fumes whether composed of liquid or solid particles, spherical

or otherwise. It is therefore somewhat remarkable as has been pointed out by Rothmund (Monats., 1918, 39, 571) that in nearly all cases the radius approximates to 5 × 10^-5 cm. It has been deduced thermodynamically by Donnan (Zeit. phys. Chem., 1903, 46, 197) that a critical radius of about 1 × 10^-5 cm. should exist, and Lewis (Syst. of Phys. Chem., 1920, 1, 332) found with oil emulsions that there was a distinct tendency for the particles to approximate to this size. But that the same radius should be found general in aerial colloids is certainly remarkable. This “critical” radius is, of course, found only when the particles have grown spontaneously and have not been formed by artificial methods such as spraying.

A very important point to be considered is the manner of production of the visible particles. It is generally agreed that some nucleus must be present whether a small solid particle (dust mote, salt crystal) or a molecular aggregate (hydrochloric acid). This is perhaps satisfactory for some cases, but does not explain the condensation of, say, ammonium chloride, unless it is granted that a few ions which may be present (and these clouds are generally uncharged) can cause the condensation of immense numbers of particles. Indeed, Aitken (Proc. Roy. Soc. Ed., 1916-17, 37, 215) has concluded that there is no evidence to show that ions of themselves ever act as nuclei—it is necessary that they be combined with an aggregate of molecules of vapour or a dust mote to form a “large ion” before condensation can occur. From a review of the literature it would appear that much more work is necessary before it will be possible to say how condensation occurs, and still more work is necessary to explain the stability of the particles, a subject which is briefly discussed in the next paragraph.

Rothmund (l.c.) has used an approximate form of the equation connecting vapour pressure with the radius of curvature of a small particle in conjunction with the van't Hoff equation for the reduction in vapour pressure of dilute solutions, obtaining finally for water particles, the equation r = 6.1 × 10^-6/ci cm. in which r is the radius of a particle with a concentration c of dissolved material, i being the van't Hoff factor. With Rothmund's ozone fogs the formula gives results of the right order. The concentration of dissolved material was very small, 0.03 to 0.08 molar. When, however, the equation is applied to the results of others and of the present author, the results are hopeless. It has been pointed out above how remarkably close to a constant radius all fog particles tend to get, although in many cases the droplets must consist of very concentrated solutions. Moreover, the equation by its derivation does not allow of the existence of a fog in a highly unsaturated atmosphere when the particles consist of dilute solutions of dissolved substances; although it has been observed repeatedly by many workers that chemical fogs can readily form and exist for long periods in such an atmosphere.

At present there is too little information in the literature to enable much progress to be made and what information there is, is rendered almost useless by the fact that in no papers known to the present author have the concentration of dissolved material in the particles and their radii, the vapour pressure of the water in the

gas and its temperature have been given together; in fact, in practically all cases only one of the above is given for the particular fog. It would appear that until all such information is given for all fogs experimented with, very little progress can be made in giving an explanation of their stability.

It is admittedly difficult to apply equations strictly to the fogs, owing to their continual change; but with some of the methods now available for the examination of clouds and smokes considerable progress should be possible.

It is a well-known fact that during the rupture of masses of materials whether solid of liquid very large quantities of electricity are generated. This is particularly so with solids (Beyersdorfer Staubexplosionen, 1925, 10, et seq.), but even with liquids it is quite considerable. It has been shown that the mere making of new surfaces is not sufficient for this purpose—the extension must be large and very rapid. Hence it may be expected that in chemical fogs charged particles might make their appearance. This expectation is to some extent borne out in practice, but generally only in such cases where violent reactions or high temperatures or other secondary methods of charging have been employed (de Broglie and Brizard C.R. 1909, 148, 1457; 149, 923). For example, in the metallic clouds formed in the electric are charged particles are present, but usually the positive and negative charges are equal in number (Whytlaw-Gray Nature, 1926, 117, 201); the vapours of phosphorus and of sulphur contain charged particles (Przibram Phys. Zeit., 1911, 12, 260); and corresponding to the third possibility given above Remy (Zeit. anorg. Chem., 1924, 139, 69) found that an uncharged fog became charged on bubbling through aqueous salt solutions, a not unexpected result in view of the work of Kösters (Ann. D. Phys., 1899, 69, 12) who found that the presence of dust or liquid particles increased the charge carried off by a gas during bubbling.

This last is the most likely source of electrical charges in the present experiments. It has been shown by a number of workers (e.g., Becker Jahr. d. Rod., 1912, 9, 52; Coehn and Mozer Ann. d. Phys., 1914, 43 1048) that gases become charged on being bubbled through salt solutions. In such a charged gas it is possible that condensation might be aided by the presence of ions (but remember Aitken's statement above). Now since with some solutions extremely dense fogs have been obtained it would be necessary to assume, if condensation were occurring on ions, that the dyes, colloids, and suspended materials conferred on the liquid the property of causing much greater charges to escape into the gas, since presumably each particle would be the result of condensation on at least one ion. In waterfall experiments some of the dyes have the property of reversing the sign of the charge on the gas, but in view of Vincent's results (Proc. Camb. Phil. Soc., 1904, 12 305) this would not be very important, if any such reversal were to occur in the case of bubbling. Other work, however, points to the above suppositions being untrue, Bloch (Ann. Chim. Phys., 1911, 23, 28) found that many substances including acetone, ethyl alcohol, amyl alcohol and hydrocarbons, when present either in solution or as layers on the surface of a liquid greatly depress or even completely inhibit the charging of a gas by bubbling.