seem impossible. With this fact clearly before us, our first work with the metals now under consideration was to ascertain how feeble a current would suffice for the complete deposition of each metal, when alone in a cyanide solution, in the presence of an excess of an alkaline cyanide. We very soon discovered that the cadmium separated readily and with a much weaker current than was necessary for the deposition of the zinc. The latter will, however, separate from a cyanide solution, even under the influence of a feeble current, but not until the excess of cyanide has been completely decomposed. With the quantity of cyanide used by us, and with a current of the strength indicated below, this complete decomposition of the alkaline cyanide is not likely to occur in a shorter period than forty-eight hours. Hence, it follows, that as the quantity of cadmium used in our experiments is entirely precipitated in a little more than eighteen hours, the complete separation of cadmium from zinc is thoroughly feasible by this method.

All our experiments were conducted in the cold, and care was taken in each case to examine the deposited metal for zinc, and the residual solution for cadmium. The conditions under which our work was carried on and the results obtained are these—

The cadmium deposit was light grey in colour and crystalline in structure. It was washed with hot water and dried upon a warm iron plate.

Chemical Laboratory of the Univ. of Pa., Philadelphia, Pa.

T me.

SOME CAUSES OF ERROR IN BLANK ANALYSES.

By J. B. MACKINTOSH.*

IN making chemical investigations, it is quite usual to run a blank analysis, in order to apply a correction to our results for the impurities contained in the reagents and for those entering into solution from the apparatus used. As a rule however, it seems to be entirely overlooked that the conditions of the actual analysis are often quite different from those obtaining in the case of the blank. For instance, let us assume the case of a limestone which has been dissolved in just sufficient hydrochloric acid to form a neutral solution of calcium chloride; at this stage our blank analysis will be a strongly acid solution. If we now add to both enough ammonia to neutralise the amount of hydrochloric acid previously employed, we will have in our actual analysis a strongly ammoniacal solution, while in our blank there will be a neutral solution of ammonium chloride, and similar results will take place on the addition of every reagent. In such a case our blank analysis will give us only approximately correct results, for though any impurities existing in the chemicals will be detected, yet the impurities which are dissolved from our beakers by the solutions, will be quite different in the case quoted, and the errors due to this last source may far exceed those due to impurities contained in the chemicals used.

The particular case of blank analysis to which I desire to call attention, serves to illustrate the serious error into which chemists may be led by faith in the infallibility of books of reference, whose statements, copied from one another, gain in authority through force of iteration, even though, as sometimes is the case, the original statement be

erroneous,

Several years ago I had occasion to analyse a large number of samples of metallic copper and alloys of copper, and for the determination of arsenic I decided to use the method of distillation with ferric chloride and hydrochloric acid recommended in Crooke's Select Methods. In that work the statement is made that the ferric chloride employed may be freed from all traces of arsenic by one or two evaporations to dryness with hydrochloric acid, whereby the arsenic will all be volatilised as chloride. In following these directions I collected the distillates, thus making a blank analysis in all respects similar to the actual analyses performed later. When the distillates were tested for arsenic they were found to be perfectly free from that element, from whence the deduction was drawn that the reagents employed were pure. If the statement mentioned above was correct, then this deduction was justified by the facts observed. I noticed, however, that all the samples of copper I had, no matter of what quality, invariably caused the appearance of large quantities of arsenic in the distillates, and on repeating the experiment with the substitution of some charcoal for the copper, I likewise obtained arsenic in the distillate, and since the charcoal could not be suspected of being the source of the arsenic found, it was definitely traced to the ferric chloride, which by a blank analysis would have been pronounced pure. In this case the omission of the copper in the blank

Journal of Analytical Chemistry.

† First Edition, p. 266; Second Edition, p. 431.

analysis was also the omission of a necessary reagent, without which or some substitute, the reaction would not take place. In fact, it is not until the arsenic is reduced to arsenious chloride that it will distil off, any arsenic in the higher form of oxidation not being volatile under these circumstances.

Here is a striking instance of the unreliability of blank analyses, unless all the circumstances be thoroughly comprehended and taken into account, and other instances of similar nature could readily be adduced.

REPORT OF RECENT RESEARCHES AND IMPROVEMENTS IN
ANALYTICAL PROCESSES.

USE OF OXYGEN IN QUANTITATIVE ANALYSIS. W. MINOR. (Zeitschr. f. angew Chemie, No. 23, 1889.-It seems strange so little use is made of oxygen in technical laboratories. Take for instance wine analyses, where the getting of a proper ash, without volatilising chlorides, is a difficult operation. If, however, the extract is heated in a current of oxygen, the ash is got ready in a few minutes. Another instance is furnished by cane-sugar and allied bodies. Although this is generally first moistened with sulphuric acid and the weight of the ash multiplied by 9, still the treatment in oxygen will dispense with the use of the acid. In the estimation of carbon in pig-iron, the use of oxygen is very advantageous. The iron is placed in solution of cupric chloride, which gradually dissolves the iron, leaving an impure carbon. This is collected on an asbestos filter dried in a platinum crucible at 120° C. and weighed. The carbon is then burnt in oxygen and found from the loss in weight. Oxygen will also be found advantageous in the estimation of sulphur or sulphates. The well-washed precipitate of barium sulphate is put moist in a platinum crucible, and after charring ignited in a current of oxygen. This will prevent all chance of reduction, and the operation is done in a few minutes. L. de K.

BORGMAN'S PROCESS FOR THE ESTIMATION OF THE ACIDS IN WINE. R. GANS. (Zeitschr. f. angew Chemie, No. 23, 1889).-This process, consisting in treating the wine extract with alcohol, to precipitate the cream of tartar and dissolve out the free acids, has been found by the author to yield uncertain results. As the results of his experiments he concludes:-1. The estimation of the cream of tartar is influenced by the amount of sugar. If much sugar is present some of the cream of tartar does not separate out, and some free tartaric acid remains undissolved. 2. The estimation of the free tartaric acid (by means of potash) is also inaccurate, as especially in absence of sufficient sugar nearly the whole of the acid potassic malate also separates out. In fact, the process can only be trusted to give approximate results in the case of very sweet wines. L. de K.

DETECTION OF SMALL QUANTITIES OF NITROUS ACID. G. LUNGE. (Zeitschr.j. angew Chemie, No. 23, 1889.)-A well-known disadvantage of the otherwise so excellent reagent of Griess for nitrous acid, is the circumstance of the solution, even of the whitest looking (a) naphtylamin, getting dark and therefore losing in sensitiveness. When the solution to be tested is very dilute (1-1,000000000), the test works so slow,

that one cannot be certain whether the nitrous acid has not been absorbed from the atmosphere. Even after warming, which facilitates the reaction, the test may not show for 15 or 20 minutes. Jlosvay shows that by using acetic instead of sulphuric or hydrochloric acid the reaction takes far less time, and the colour is also intensified. He also states that the darkening of the reagent may be prevented by boiling some solid naphtylamin with water, and only using the fluid which has been poured off from the residue. In fact, he gives the following recipe:-1. Dissolve 5 gramme of sulphanilic acid in 150 c.c. of dilute acetic acid. 2. Boil 1 gramme of solid (a) naphtylamin with 20 c.c. of water, pour off from the blueish deposit, and mix with 180 c.c. of dilute acetic acid. The liquid to be examined (about 20 c.c.) is mixed with a few c.c. of the sulphanilic acid, warmed till about 80° C., and then mixed with the solution of the naphtylamin. Even when the solution contains only one thousand of a millionth part of nitrous acid, the red colour will appear after one minute. If the proportion of the nitrous acid is larger, say 1 in 1000, only a yellow colour will appear, unless a large quantity of naphtylamin is added. The author on the whole agrees with Jlosvay, only thinks he can still further improve it. The test for nitrous acid is just as delicate and quick, if two reagents are mixed and kept in a well-stoppered bottle. Air must be as much as possible excluded, but it is not necessary to keep the reagents in the dark. By this small modification we get the great advantage of using one solution only. Then if the two solutions are kept separate, one can never tell whether perhaps one of them has absorbed a little nitrous acid from the air. If, however, the mixed solution should absorb any nitrous fume, it would at once betray this by the reddish colour. If a solution should have got coloured, it is often possible to restore it to its original condition by shaking it with zinc dust and filtering. L. DE K.

E. GLASER. (Zeitschr. f.

ESTIMATION OF IRON AND ALUMINA IN PHOSPHATES. angew Chemie, No. 22, 1889.)-The process for the estimation of iron and alumina in phosphates, as now practised, is not applicable to any variety of phosphate, and therefore not very useful. Its shortcomings have shown themselves more particularly in the last few years, when many samples of Belgian and similar phosphates are on the market. The results are somewhat dependent on the amount of acetic acid used, then aluminic phosphate is somewhat soluble in excess. This conventional process is as follows: The precipitate insoluble in acetic acid is filtered off, washed with cold water, and then dissolved off the filter with hydrochloric acid. After adding a little microcosmic salt the liquid is rendered alkaline with ammonia and then mixed with acetic acid. The precipitate is washed, dried, and ignited, and half of its weight is called iron and alumina. No wonder analysts differ in their results from 1 to 3 per cent. Not only, however, is the error solely due to the slight solubility in acetic acid, but also to the fact that in the original sample, the iron is not always in combination with phosphoric acid. When such a solution is precipitated with ammonia, the lime and magnesia are precipitated as phosphates, which is also the case with the iron and alumina, if there is a sufficiency of phosphoric acid. This, however, is not always so, and then there forms a precipitate of ferric and aluminic hydrate, which redissolves in excess of acetic acid and so get lost.

Then again the ferric acetate has a solvent action on ferric phosphate. If too little acetic acid is used the precipitate may contain lime and the result is equally faulty.

(Note by abstractor: Iron and alumina phosphates will not properly come down with ammonia and acetic acid if the phosphate contains organic matter, which is very often the case. This being absent, all iron and alumina will certainly come down, except in cases where there is a want of phosphoric acid).

The author now tried to find a process which will dispense with the acetic acid, and succeeded by the use of alcohol as follows: 5 grammes of phosphate are dissolved as usual in 25 c.c. of nitric acid of 1.2 sp. gr., and 12·5 c.c. of hydrochloric acid, and diluted up to 500 c.c. 100 c.c. of filtrate (= 1 grm. sample) are put into a 250 c.c. measure and mixed with 25 c.c. of strong sulphuric acid. After standing for about five minutes and occasional shaking, 100 c.c. of alcohol are added, and the whole well shaken. After cooling alcohol is added up to the mark, but as there is always a little contraction it will be found necessary to later on add some more spirit. After at least standing for half an hour the liquid is filtered, and 100 c.c. (= 4 grm. sample) are evaporated in a platinum dish till alcohol is expelled. The acid fluid is rinsed into a beaker with about 50 c.c. of water, and heated to boiling. After cooling a little ammonia is added to alkaline reaction, and if necessary the excess of ammonia is expelled by boiling. The precipitate is now collected, well washed, dried, and ignited. It consists of pure ferric phosphate and pure aluminic phosphate. The iron phosphate can be determined by a volumetric estimation, say with permanganate.

The test analyses given are very satisfactory, more so than the analyses performed by the conventional process. L. de K.

DETECTION OF GROUND-NUT OIL IN COD-LIVER OIL. V. BISHOP. Journ. de ph, et de chim, xx. p. 302.-Ground-nut oil is frequently added to cod-liver oil with a double object, viz., to give to the Cod-liver oil a more pleasant taste and to increase the profits of the vendor. It is therefore to our great interest to be able to detect it, and seeing that it does not possess characteristic properties, as does, for instance, oil of sesame, its detection is verydelicate and uncertain. Nevertheless, having studied the question the author believes it to be quite possible to arrive at a satisfactory solution. The colouration furnished by acids-notably by sulphuric acid, the estimation of the iodine, etc., do not always give exact indications. The density, the heating power of the sulphuric acid, the absorption of bromine are three modes of analysis capable of giving much more exact results, on account of the notable differences which exist in these respects between the two oils. To reply, then, the author works in the following manner :

1st. The density is taken by the areothermic balance with the usual precautions. 2nd. The action of sulphuric acid on cod-liver oil and even on ground-nut oil being very energetic, and producing a strong disengagement of sulphurous acid, it is indispensable to mix them with a known quantity of, a retarding oil, the heating power of which is at once low and ascertained: With this object the author employs heavy mineral oil, which, as is well known, possesses a very low point of heating power, with sulphuric acid, and such oil he mixes with the cod-liver-oil in equal proportions.

3rd. In applying the bromine absorption method, the author employs M. Halphen's process, by which the oil is saponified and the fatty acids liberated in the usual manner. 1 c.c. of these acids having been dissolved in 20 c.c. of carbon bisulphide, the solution is treated with a fixed excess of a standardised solution of brominę,