These results bear two explanations:-(1) The metal is volatile per se at these high temperatures, or (2) the metal unites with oxygen at a certain temperature, forming a volatile oxide, which may either volatilise unchanged, or may decompose at a higher temperature with deposition of the metal, the oxygen acting in this case as the temporary carrier of the metal. It is probable that each of these causes is active in one or other of the metals experimented with. I will take the case of the metal first tried, viz., platinum. The universal experience of chemists is against the supposition that platinum forms a volatile oxide below 1300°. Deville and Debray, who worked long and thoroughly on the platinum metals, agree that" platinum is distinguished from all the metals which accompany it in the mineral, by the fact that it does not unite direct with oxygen in whatever condition the two bodies are placed." (Comptes Rendus, 1878, vol. lxxxvii., p. 441).

The mode of occurrence of the beautiful crystals of platinum is against the supposition that they are a product of the decomposition of an oxide, for the crystals deposit on a part of the apparatus that is at a slightly lower temperature than the bulk of the metal, and it is inconceivable that platinum should unite with oxygen to form a volatile oxide at one definite temperature, and part with this oxygen and come down in metallic crystals at a little lower temperature.

The boiling point of platinum is put by Kaye and Laby at about 2450°(Physical Constants," 1911, p. 49, Longmans). Moissan's electric furnace in which the arc is used is said by him to give with ease a temperature of 3500°, and here platinum enters into fusion in a few minutes, and soon volatilises. The metallic platinum collects in small brilliant globules and in powder on the cooler parts of the electrodes, or on the surface of the lower brick some centimetres from the crucible (Moissan, "Le Four Electrique," p. 43). No mention is made of the metal condensing in crystals.

I must therefore come to the conclusion that platinum is absolutely non-volatile at 900°, a temperature easily attain able in an analytical laboratory, and that the formation of crystals of the metal in the electric furnace is a true case of sublimation.

The next metal, iridium, under experiment behaves differently to platinum. It is admitted that iridium is volatile at a high temperature. Thus Mendenhall and Ingersoll (Physical Review, xxv., 13) speak of melting iridium into a globule and then remelting it two or three times before it entirely sublimed. This at a temperature above its

melting-point, which these authors put at between 2300° and 2400°.

Waidner and Burgess (Bulletin of the Bureau of Standards, Washington, vol. iii., p. 183) describing an "iridium furnace," of which the essential part is an iridium tube 25 cm. long, 2 cm. in diameter, and 0.25 mm. wall thickness, state that to avoid evaporation of the iridium it was coated with Nernst's refractory earth mixture.

These illustrations, however, are taken from observations of experimentalists who were working with iridium at or near its melting-point.

At high temperatures below 1000° iridium oxidises, forming an oxide, Ir2O3, which is volatile. Deville and Debray consider the alleged volatility of the metal is due to the formation of this volatile oxide. They say (Comptes Rendus, lxxxvii., p. 445), "Above 1000° all volatilisation becomes impossible in our atmosphere, because the oxide of iridium ceases to exist, and the metal is at least as fixed as platinum."

In my experiments the cold iridium was put into the furnace previously heated to 1300° or thereabouts, and the metal quickly became hot. While it was rising to 1000° accounting for the loss of weight. it oxidised superficially, and the oxide rapidly volatilised,

I devised an experiment to see if iridium would volatilise at a high temperature in a vacuum.

A fused silica tube, 1 cm. diameter and 20 cm. long, had a bulb 2 cm. diameter blown on the end (Fig. 10). In the bulb was put 27.619 grains of clean iridium, and the other

end of the silica tube was drawn out for connecting with the pump and sealing. It was exhausted to a high vacuum and heated to near redness along its whole length till all moisture and occluded gases had been removed; it was then sealed off.

The tube was placed bulb uppermost in the furnace in such a position that the iridium would be at the point of greatest heat, the lower part of the tube remaining com. paratively cool. The bulb was kept at a temperature of 1300° for thirty hours. On examining the silica tube when cold it was seen that the long continued high temperature had caused the bulb and the upper part of the tube to devitrify, and become white and translucent and that it had an irregular black deposit on the lower part, which, on examination, proved to be metallic iridium.

The iridium removed from the bulb was now found to weigh 27.600 grains, showing a total loss of oo19 grain or o 069 per cent in thirty hours.

The presence of a deposit of iridium in the cooler end of the silica tube may thus be accounted for; during the heating of the bulb containing the iridium a little air

CHEMICAL NEWS

May 24, 1912

243

absence of nitrites, no action was produced on the starchiodide reagent. Thus a solution containing ten parts of ferrous iron per million, and no nitrite, did not give the

I must not conclude this paper without expressing my thanks to the firm of Johnson and Matthey, to whom I am indebted for many specimens of the pure metals used in this research.

iron and nitrite were present together in solution it was found that the iron retarded the production of the blue colour by the nitrite, and therefore the nitrite was underestimated.

From these and similar experiments it was found that ferrous iron in quantities exceeding o'1 per million causes a diminution in the blue colour produced by nitrites with the starch-iodide reagent. A few of the results are given

By G. D. ELSDON, B.Sc., A.I.C.

In a well-known English text-book on water analysis the following sentence occurs in connection with the estimation of nitrites by the starch-iodide method :-" Ferric salts also liberate iodine from potassium iodide when present in any quantity, but no potable water ever contains sufficient to produce the reaction."

Some little time ago the author had reason to doubt this statement on account of some peculiar results which he obtained when using the starch-iodide reagent, and some experiments were undertaken with a view to finding out what dilution of ferric iron would produce the "nitrite " reaction, when tested with starch and potassium iodide. In order to do this standards were made up of various strengths of ferric iron, and these were then tested with starch and potassium iodide, the colours produced being matched with those obtained from solutions of sodium

nitrite of known strengths treated in a similar manner. The method used was as follows:-Fifty cc. of the water to be tested were put into a Nessler glass and I cc. of roughly normal hydrochloric acid added, followed by I cc. of starch solution (0.5 per cent), and I cc. of potassium iodide solution (16 per cent)-it being necessary to observe this order. Proceeding in this way the following

results were obtained :

Time for formation of colour.

30 seconds

:

Parts of ferric iron.*

ΙΟ

Parts of ferrous iron. Parts nitrite present. Parts nitrite found.

The effects of ferrous and ferric iron on the Griess

Ilosvay test were tried both in the presence and absence of nitrites. It was found that quantities of ferric iron varying from o'or to 15 parts per million produced no colour with the reagent, and, further, no colour was produced by ferrous iron in strengths varying from one to twenty parts per million. It was found also that ferrous and ferric iron did not affect the colour produced by nitrites. Fifty cc. of the water are placed in a Nessler glass, and The Griess-Ilosvay test is carried out as follows:2 cc. of the reagent added. The colour produced is matched against similar glasses containing known amounts of nitrite. The reagent is made by heating o'i grm. of a-naphthylamine with 20 cc. of glacial acetic acid, and when dissolved mixing the resulting solution with 130 cc. of roughly normal acetic acid; to this is then added o'5 grm. of sulphanilic acid dissolved in 150 cc. of roughly normal acetic acid.

Conclusions.

For the following reasons the author much prefers the Griess-Ilosvay reagent for testing for nitrites, both quali tatively and quantitatively :

1. The colour is practically steady after standing about two hours, and it is easy to compare and match. 2. It is not affected by ferrous and ferric salts.

3. Only one solution has to be added, and this will keep for months without decomposition.

The starch-iodide test only seems to have one good point, namely, that the solutions required are ordinary laboratory reagents.

The Significance of Nitrites in Potable Waters.

The authority cited at the beginning of this paper states that :-"Water containing a trace, however slight (of nitrite), must be regarded with grave suspicion unless some other source is found from which they have more probably been derived." It is quite obvious that the value of this remark depends entirely on the amount of nitrites which can be called "the slightest trace." The GriessIlosvay test is capable, with care, of detecting one part of nitrous nitrogen in a thousand millions of water (or even less), and no water has come under the author's notice which has not given a trace of colour with this reagent after standing four hours, not excluding several public supplies to which not the slightest suspicion can be attached. Few satisfactory waters have contained more than 0.005 parts of nitrous nitrogen per million, and none more than o'or per million. It seems probable, therefore, that a water would be entirely free from suspicion, as far as nitrites were concerned, if it contained less than o'or part of nitrous nitrogen per million.

In use the T-piece c is approximately adjusted and then the rod a moved up and down for fine adjustment. Actual records of 215 observations gave an average temperature of 45'04° C. to.ro for 45° C. intended.

(We are indebted to the Author for the woodcuts illustrating this Paper).

PROCEEDINGS OF SOCIETIES.

ROYAL SOCIETY.
Ordinary Meeting, May 9th, 1912.

Sir ARCHIBALD GEIKIE, K.C.B., President, in the Chair.

PAPERS were read as follows:

"Variation with Temperature of the Rate of a Chemical Change." By A. VERNON HARCOURT, F.R.S.

In an inquiry into the connection between the conditions of a chemical change and its amount, one of the conditions varied was that of the temperature of the solution in which the change took place (Phil. Trans., 1895, clxxxvi., A, 817-895). A relation was found to exist between this condition and the rate of change, expressed by the equation ar/ar (T/To)m, where a is the rate of change, or the number of minutes in which a definite portion of chemical change is accomplished, To the absolute temperature 273°, and T any other absolute temperature.

Not only do the numbers found from this equation agree very closely with the observed numbers, but the equation expresses a natural law which is nearly related to that upon which all calculations of gaseous volumes have long been based. For as a highly probable extrapolation shows that at -273° C. the tension of gases falls to zero, and it is inferred that molecules cease to move, so it appears from these observations that this same figure represents the highest temperature at which chemical change does not take place, and atoms cease to move. At this temperature silicon and fluorine, potassium, and nitric acid, would lie down together; silver would not tarnish nor iron rust.

Several later measurements of the rate of change at different temperatures have been published and compared with numbers calculated from other formulæ. In an appendix to the present paper it is shown, by one of the authors of the previous paper, that the numbers thus calculated are in less close agreement with the actual measurements than numbers calculated from his formula given above, while also the formulæ have no physical interpretation. His colleague has, at the same time, made some observations on another case of chemical change, of which the following is a summary :

When stannous chloride is added to the deep red liquid made by mixing solutions of ferric chloride, hydrogen chloride, and potassium sulphocyanide, the ferric salt is gradually reduced, and the red colour fades. The warmer the liquid is the more quickly does this happen. Observations of the rate of change were made by mixing these substances in a glass cylinder, by the side of which was placed a similar cylinder containing the same volume of a mixture of solutions of potassium sulphocyanide and hydrogen chloride with a smaller quantity of ferric chloride. This cylinder and its contents were kept as a standard, corked and in the dark when not in use, deep red but translucent. A second much paler standard was made up similarly, but with only two or three drops of the ferric chloride solution. A few minutes after the admixture with stannous chloride the liquid in the reagent cylinder, which at first is nearly black and opaque, approached in appearance the dark standard. The observer, watch in hand, looked at the two at intervals of about two seconds, and noted the

When the colour had nearly faded to that of the pale standard, a second comparison was made, and the time noted. The interval between the two observations was the time in which, at each chosen temperature, one and the same piece of chemical work had been done. Such a pair of observations was made at every three degrees from 9 to 30, and of such sets of observations three were made under very nearly the same conditions. The times given below are the mean of the three sets, and those below them represent rates calculated from the equation ar=0.00845(T/To)28 5, being their reciprocals, or the times in which the piece of chemical work would be done at those rates. Temps. 12° 15° 18° 21° 24° 27° 30° Time (mins.) 472 349 25'9 19.1 14'6 108 80 6'1 Calculated.. 472 349 25'9 193 14'4 10.8 8.1 6.1

9°

When the proportion of hydrogen chloride is increased the rate also increases. A set of observations was made over the same series of temperatures in which the intervals were about one-third of those above. The observed and

calculated times were in almost as good agreement.

An addition of hydrogen sulphate diminishes the rate; an addition of hydrogen nitrate leaves it nearly, if not The cause of the former may be that quite, unaltered. ferric sulphate is less easily reducible than ferric chloride. An increase of potassium sulphocyanide increases the rate, and so does an addition of sodium chloride.

The work that has been done is suggestive of much further work on the same lines, and the author hopes that it may now attract the interest, and pass into the hands, of some younger chemist, to whom he would gladly give any help which his experience of such work might enable him to give.

"Some Phenomena of Sunspots and of Terrestrial Magnetism at Kew Observatory." By C. CHREE, SC.D., LL.D., F.R.S., Superintendent, Kew Observatory. An investigation made some years ago by the author indicated the probability that a relation existed between the amplitude of the daily range of the magnetic elements and the sunspot area, not on the same day but several days previously. The object of the present research was to inquire into the reality of this connection. A selection was made of the five days of each month of the eleven years 1890 to 1900 which had the largest sunspot areas as given by the Greenwich annual lists. Mean values of the sunspot areas were derived for the 650 days thus selected (two months were omitted as having less than five days showing any sunspots), and for thirty other groups of days of the same number, corresponding to the fifteen days immediately preceding and the fifteen days immediately succeeding each of the 650 selected days. In this way one got thirty-one representative successive days, of which the central day had about twice as large a sunspot area as the average. The sunspot area rapidly and regularly declined on either side of the central day to an almost dead level, thus giving a very prominent "pulse" of sunspot area. The Kew daily horizontal force ranges were got out for the 650 representative days of large sunspot area, and the allied 19,500 days, and mean values obtained again for the thirtyone representative days. These mean values gave a marked pulse, corresponding to the sunspot area pulse, but with its crest about four days later. They gave also a minor or secondary pulse about fifteen days prior to the principal pulse. Several attempts were made to arrive at the cause of the secondary pulse. It was found to be largely a disturbance effect. In investigating its nature, it was found that there is a well-marked period of about 27.3 days in magnetic phenomena, in this sense, that if a certain day exhibits magnetic disturbance attaining the international standard "2," as interpreted at Kew, a day which follows either twenty-seven or twenty-eight days after has nearly double the chance of attaining standard