of the formation of zinc iodide and zinc bromide, which pass into the water; and we have thus an aqueous solution of the two latter salts above, and colourless carbon disulphide at the bottom. The latter is drawn off by a syphon, and restored to the kelp solution, to which some more chlorine water is added; and, if a further quantity of iodine and bromine be liberated, the above operations are repeated till the liquor is quite exhausted.

It only now remains to evaporate the solution of zinc salts, divide into 3 equal portions, and determine iodine, bromine, and chlorine, as before described.

Detection of Bromides in the Presence of Chlorides.

When gold chloride is added to a faintly acid solution of an alkaline bromide, a colouration is produced ranging from dark orange red to light straw colour according to the strength of the solution. Iodides must not be present; chlorides, however, do not interfere. The best mode of proceeding is as follows:-Remove the iodides, if present, by means of palladium, and after getting rid of excess of palladium by sulphuretted hydrogen, concentrate the solution to about 25 c.c. Select two test-tubes of the same size and shape and colour of glass; into one pour the solution suspected to contain bromine, and into the other pour pure water containing a trace of potassium chloride. Add to each tube one drop of hydrochloric acid, and one drop of gold chloride solution. On now comparing the two tubes, particularly in the direction of the long axes, a yellow colour will be observed in the tube containing the bromide, which will be rendered very manifest by comparison with the other tube.

The mixed chloride and bromide should be brought to the state of alkaline salts if necessary, by precipitating with silver nitrate, thoroughly washing and fusing with potassium carbonate. If sodium carbonate is used for this purpose the subsequent reaction with the gold test is not so decided.

Detection of Chloride in Potassium Bromide.

The bromide to be examined is first tested for iodine. For this purpose a small quantity of the salt is dissolved in water in a testtube, and an equal volume of carbon disulphide added. Upon the addition of a few drops of bromine water, the carbon disulphide becomes coloured violet, under the influence of iodine, if this be present. When the test shows the presence of iodine, it is necessary to remove the whole of this element from the sample. This is effected by dissolving about 10 grammes of the salt in distilled water, adding bromine water until violet vapours are no longer visible upon boiling, and then testing for iodine in the manner first described. Afterwards the solution is evaporated to dryness to remove the excess of bromine,

and thus is obtained a potassium bromide free from iodide, but which may contain chloride.

The remainder of the process depends upon the fact that a given weight of potassium chloride requires, for complete precipitation, a much greater amount of a standard solution of silver nitrate than the same weight of potassium bromide. While the bromide for the complete precipitation of 1 gramme requires 1.428 gramme of silver nitrate, 1 gramme of the chloride requires 2.278 grammes. A standard solution of silver nitrate is first prepared by dissolving 10 grammes of the pure salt in a litre of water, each c.c. corresponding to 1 milligramme of silver nitrate; 1 gramme of the bromide to be examined, freed as above from iodine, is dissolved in 100 c.c. of distilled water; 10 c.c. of this solution, representing 0.1 gramme of potassium bromide, would require, if pure, 14.2 c.c. of the silver solution; potassium chloride would require 22.7 c.c.

M. Baudrimont has proposed a method of making the final reaction more delicate, by adding a few drops of solution of potassium chromate to the bromide examined; the silver nitrate added at first combines with the whole of the bromine and chlorine in preference, and the complete precipitation is marked by the production of the red precipitate of silver chromate. It is obvious that the bromide contains more or less chloride, according as the number of burette divisions (divided into c.c.) of the silver salt required exceeds 142. With a salt containing of its weight of potassium chloride 151 divisions are required, and with a mixture of equal weights of chloride and bromide, 185.

The same method may be employed to recognise the degree of purity of several compounds. Operating as before-that is to say, dissolving 1 gramme of the material to be examined in 100 c.c. of distilled water, and taking 10 c.c. of the solution-the following numbers of

c.c. divisions required will show the purity for at least a considerable number of salts:-102 for pure potassium iodide, 257 for potassium cyanide, 246 for dry potassium carbonate, 290 for sodium chloride, 119 for sodium carbonate + 10 equivalents of water, 47 for sodium phosphate + 24 equivalents of water, and 54 for sodium arseniate 14 equivalents of water.

Detection of Iodine in Potassium Bromide.

When potassium bromide is suspected to be adulterated, or mixed with potassium iodide, place a few grains of the salt in question on paper previously impregnated with starch-paste, moisten it, and admit a small quantity of chlorine gas, whereby the iodine is set free and the paper coloured blue.

A better test is the use of bromine water added to the salt after it has been placed in benzol; if the latter becomes red-coloured, iodine is present.

CHLORINE.

Estimation of Chlorine with the aid of Gooch's Method of Filtration.

Mr. David Lindo remarks that it is generally considered that chlorine can be estimated with great exactness by the gravimetric method. Silver chloride being slightly soluble in water, especially in hot water, a small minus error may occur if the latter is employed to wash with, but this can be prevented by adding a little silver nitrate to the water, as recommended by J. P. Cooke.1

On the other hand, the precipitate retains occluded matters with great force. Error from not completely removing these often more than compensates for slight loss occasioned by the use of hot water alone, or merely acidulated with nitric acid, or by the manipulations necessary when paper filters are employed.

2

According to Fresenius, we can, with great care, always obtain by this method 99.9 to 1001 for 100 parts of chlorine taken. I presume Fresenius means when using paper filters, in which case the time required to make an estimate is generally six hours.

The limits of error here laid down are often reached, according to my experience, when paper filters are employed, and no silver nitrate added to the wash water.

Though sufficiently near for most purposes, greater accuracy chlorine estimates may sometimes be desired. By adopting Gooch's method of filtration and Cooke's suggestion, with a few other simple precautions, a much higher degree of accuracy can be attained with less manipulation and expenditure of time than by the usual method.

Method:-Weigh the solution in a light glass stoppered bottle, and turn it into a deep porcelain capsule, about 4 ounces capacity, provided with a well-formed lip and handle. Rinse the bottle with 25 c.c. distilled water. Add solution of silver nitrate in about the proportion of 25 c.c. to 0.5 gramme potassium chloride, and 2 c.c. pure nitric acid, sp. gr. 1.2. Heat to boiling-point, and keep at this temperature for some minutes without allowing violent ebullition, and with constant stirring, until the precipitate assumes the granular form. Allow to cool somewhat, and then pass the fluid through the asbestos. Wash the precipitate by decantation with 200 c.c. of very hot water, to which has been added 8 c.c. nitric acid and 2 c.c. dilute solution of silver nitrate containing 1 gramme of the salt in 100 c.c. of water. The washing by decantation is performed by adding the hot mixture in small quantities at a time, and beating up the precipitate well with a thin glass rod after each addition. The pump is kept in action all

Chemical News, vol. xliv., p. 235.

2 Quantitative Analysis, Seventh Edition, p. 170.

the time, but to keep out dust during the washing the cover is only removed from the crucible when the fluid is to be added.

Put the capsule and precipitate aside, return the washings once through the asbestos so as to obtain them quite clear, remove them from the filter, and set aside to recover excess of silver. Rinse the receiver and complete the washing of the precipitate with about 200 c.c. of cold water. Half of this is used to wash by decantation, and the remainder to transfer the precipitate to the crucible with the aid of a trimmed feather. Finish washing in the crucible, the lumps of silver chloride being broken down with the glass rod. Remove the second filtrate from the receiver, and pass about 20 c.c. of alcohol at 98 per cent. through the precipitate. Dry at 140° to 150°. Exposure for half-an-hour is found more than sufficient, at this temperature, to dry the precipitate thoroughly.

In the volumetric estimation of chlorine, with a standard solution of silver and potassium chromate as an indicator, Prof. A. R. Leeds points out that the chromate employed is more or less contaminated with alkaline chlorides. Hence it is necessary to estimate the number of tenths of a c.c. of the silver solution which correspond to the number of drops used of the particular chromate solution.

Detection and Estimation of Chlorine in presence of
Bromine and Iodine.

G. Vortmann has discovered a method by means of which even small quantities of chlorine along with the other halogens can be easily and quickly detected. It depends on the different behaviour of the chlorides, bromides, and iodides with manganese and lead peroxides in presence of acetic acid.

Iodides are partially decomposed by the above-mentioned peroxides, even in neutral solutions, and if they are boiled with the addition of acetic acid the iodine is completely eliminated. Lead peroxide oxidises a part of the iodine to iodic acid, but with manganese peroxide no iodic acid is formed.

In a neutral solution bromides are not decomposed either by manganese or lead peroxide. In an acetic solution the lead peroxide only acts; bromine escapes; but bromic acid is formed only if bromides are present in considerable quantities. Manganese peroxide has no action in the acetic solution, even on prolonged heating.

Chlorides are not attacked by either of the peroxides in presence of acetic acid. In testing for chlorides in presence of bromides or iodides it is sufficient to boil the substance in an acetic solution with lead peroxide till the liquid on settling is colourless and has not the slightest odour of bromine or iodine. The bromine and a part of the iodine escape as such; the remainder of the iodine remains as lead iodate along with the excess of the lead peroxide. On filtering and washing the precipitate, all the chlorine is found in the filtrate free

from bromine and iodine. In this manner the chlorine may be estimated quantitatively. If the quantity of chlorine accompanying the iodine is considerable, manganese peroxide is preferable to lead peroxide, as otherwise the liquid must be largely diluted with water to prevent lead chloride from depositing. In estimating large quantities of chlorine in presence of bromine, it is well to add along with the lead peroxide some potassium sulphate so that all the chlorine may be found in the filtrate combined with potassium.

In order to expel the liberated bromine and iodine more rapidly a moderate current of air may be passed through the solution on the water-bath.

Detection and Estimation of Iodine in presence of
Bromine and Chlorine.

E. Donath remarks that whilst potassium bichromate occasions in solutions of potassium iodide no separation of iodine, this result is at once produced by chromic acid. An estimation of the liberated iodine is not possible on account of the action of the excess of chromic acid upon sodium thiosulphate. The iodide of starch is precipitated by solutions of chromic acid as an almost black precipitate. Hence, starchpaste cannot serve as an indicator. The iodine is therefore distilled off, and is estimated in the distillate by means of sodium thiosulphate. The alkaline bromides and chlorides are not decomposed by concentrated solutions of chromic acid at common temperatures, and the chlorides not even on boiling. Dilute solutions of chromic acid decompose the bromides only to a very small extent at the boiling heat, but the proportion increases with increasing concentration. The method is therefore suitable for accurate estimations of iodine in presence of chlorine, but not when it occurs along with bromine. The solution of chromic acid employed contains from 2 to 3 per cent., and is freed from traces of sulphuric acid by boiling with pure barium chromate.

Estimation of Chlorine in Bleaching Powder.

(a) The commercial estimation of bleaching powder only extends. to the estimation of the hypochlorite contained therein; the result being, however, calculated as so much per cent. of 'available chlorine.' Of the numerous methods proposed for the estimation of the hypochlorite, the one usually employed in the trade is that depending on the amount of ferrous salt oxidised by a given weight of bleaching powder. It frequently happens, however, that instead of a perfectly pure ferrous salt (such as the ammoniosulphate precipitated by alcohol), the ordinary iron protosulphate of the druggists is used, discoloured crystals being of course rejected. This substance is, however, rarely pure, and hence errors are frequently introduced, less chlorine being required to peroxidise a given weight of impure than of