To overcome this diffiulty a second form of polariser has been devised in which three total reflections are used. The bi-trapezoidal form of the section is shown in Fig. 2. If, as before, μdi 15035, and the acute angle of the trapezoid be 73° 48'-6, then the total phase difference inA

=

This form of half-shade-using white light—has been compared with the Laurent half-shade-using monochromatic light-and was found to be quite as efficient as the Laurent. Since the cost of large blocks of calcite is becoming more prohibitive to their use, this form of white

В

troduced between the components polarised parallel and perpendicular to the plane of incidence, for the three reflections indicated, is π/2. The ellipticity of the transmitted light is smaller for a given wave-length than it is when a Fresnel's rhomb is used, and so this form is more efficient than Fresnel's rhomb. Moreover, this form possesses the advantage of compactness.

In the B.A. Report for 1851, Prof. G. Stokes claims a high degree of accuracy to be attainable by his elliptic analyser, but since it is somewhat difficult to make accurate settings of a quarter wave-plate and the analysing Nicol's prism, it was thought this second form of polariser would prove useful in the analysis of elliptic vibrations. One axis of the quarter wave-plate would be replaced by an edge of the glass, which should be specially worked, and accurate settings of this edge with respect to the analysing Nicol's prism could be easily made.

Owing to the fact that light polarised parallel to the plane of incidence is more copiously reflected on crossing the joint than the perpendicular component is, the trans

THE arrangement of rhombs described in the first form of polariser may clearly be used to form a white light halfplate. Taking the refractive index of the glass as 1'5173, and the angle of the rhombs 55° 15'5', the four total reflections produce a phase difference between the components amounting to π. The two blocks of glass C and D (Fig. 1) partly compensate for loss of light by absorption, but their chief use is to compensate for loss of light at the four reflections at normal incidence.

Half the field of view is occupied by the aperture of a rhomb, and the other half by either of the blocks c or D. By making the edge of the rhomb overlap the block slightly, the line of separation of the two halves of the field is made quite sharp.

A RAPID PROCESS FOR THE

ESTIMATION OF COCOANUT-OIL IN ADMIXTURE WITH BUTTER-FAT.

By NOEL C. CASSAL, A.R.C.S., B.Sc. (Lond.),

and

B. HENRY GERRANS, F.I.C.

THE determination of cocoanut oil when in admixture with butter-fat presents special difficulties, and a large number of processes for the estimation of the cocoanut-oil have been put forward, both in this country and abroad. Of the various methods, the most satisfactory are those which are based upon a consideration of the proportion of insoluble volatile fatty acids yielded on distillation under determined conditions. The titre of the insoluble volatile fatty acids obtained from cocoanut-oil is higher than that of the insoluble volatile fatty acids from butter-fat, but the extent of the difference shown in this respect is dependeut entirely upon the conditions of experiment.

Thus in Wauter's well-known method of double distillatior. (one of the best methods at present known, both from a manipulative and quantitative point of view), working on 5 grms. of filtered fat, a titration figure of 15'4 (cc. N/10 KHO) is obtained for the insoluble volatile fatty acids from cocoanut-oil, and a titration figure of o'g (cc. N/10 KHO) for the insoluble volatile fatty acids from butter-fat-a total difference of 14.5. While the difference is but small when mixtures are under examination, the low figure for butter recommends the process.

Polenske's process gives a difference between the titration figures for insoluble volatile fatty acids, which is small, and the determination of the cocoanut-oil is based upon a consideration of the relationship existing between the titration figures for insoluble volatile and soluble volatile fatty acids, and upon the appearance of the acids which distil over. Polenske also works with 5 grms. of fat, The

process is highly empirical, and a special apparatus of standard dimensions is required.

In the process described below, the present authors have succeeded in markedly increasing the difference between the titration figure for the insoluble volatile fatty acids from cocoanut-oil, and the titration figure for the insoluble volatile fatty acids from butter. The process is as follows:

Three grms. of the filtered fat are saponified in a 300 cc. flask with 10 cc. of absolute alcohol, and 2 cc. of aqueous sodium hydrate (50 grms. per 100 cc.) The alcohol is driven off, the residual soap is dissolved in 50 cc. of boiling water, 10 cc. of concentrated hydrochloric acid is added, and then 50 grms. of anhydrous granulated calcium chloride. The calcium chloride must be added after the liberation of the fatty acids by the hydrochloric acid. The flask is then fitted with a cork through which pass two glass tubes. One of these is about 7 mm. in diameter, and is bent at the centre at an angle slightly greater than 90°. The free end of this tube passes through a rubber-stopper which fits into the wide end of an ordinary Liebig condenser. (A Liebig condenser 49 cm. in length is a convenient size). The other glass tube is bent in that length which is inside the flask along the neck and down the concave side of the flask, and is slightly turned up at the tip. The tip is pointing away from the condenser, and almost touches the bottom of the flask. The free end of the tube, which is 5 mm. in diameter, is straight, and is connected by a rubber piece to a boiler. The flask is immersed down to the level of the contained liquid in a calcium chloride bath, boiling briskly but not violently at 141-146° C. The flask is inclined in the opposite direction to that which is usual in steam distillations (i.e., with the neck pointing towards the condenser) in order to avoid an acute angle in the exit tube. When ebullition commences in the flask the steam from the boiler is turned on, and the steam distillation is proceeded with until 500 cc. of distillate have passed over. The distillate is collected through an ordinary wet English filter into a 500 cc. flask. The passage of the steam must not be so rapid as to cause the free end of the inner condenser tube to become hot. The calcium chloride bath is kept at 141-146° C. by the cautious addition of cold water when required. The distillation occupies from one and a-half to one and threequarter hours, and requires practically no attention except the occasional addition of water to the chloride bath.

When 500 cc. of distillate have been collected the tube conveying the steam to the flask is quickly disconnected from the boiler, and then the exit tube from the flask is disconnected from the condenser. The liquid in the 500 cc. flask is rejected.

The inner condenser tube is well washed out several times with cold water on to the filter, and the filter and the insoluble acids are then well washed with cold water to

remove hydrochloric acid and any soluble fatty acids. All these aqueous washings are rejected. The condenser tube is then washed out with hot methylated spirit into a small flask, and finally the acids on the filter are dissolved out with hot methylated spirit, the solution being added to the alcoholic condenser washings. The solution thus obtained of the insoluble volatile fatty acids is then titrated with N/10 baryta.

The dimensions which have been given for parts of the apparatus employed are set forth as a guide and for the sake of convenience, but, with the exception of the size of the distilling flask, are not to be regarded as rigidly essential. The general arrangement of the parts of the apparatus should, however, be adhered to. The following

results were obtained :

The results which are given in the table have not been specially selected. Those mixtures of the same percentage composition which are marked with an asterisk were made up with different specimens of butter-fat, and are not duplicate results for one and the same mixture. The figures given for the mixture containing 35 per cent cocoanut-oil represent the results of duplicate experiments with the same butter-fat.

The appearance of the fatty acids which distil over is of interest. In the case of butter-fat towards the end of the distillation, the acids coming over are in very bulky woollylike masses which solidify in the condenser tube and on the filter. The same phenomenon is presented in the case of cocoanut-oil, although not so markedly, at a considerably later stage of distillation. In mixtures of cocoanut-oil and butter-fat the bulk of these solid fatty acids is much reduced.

The calculation of the results is simple. If A represents the titration figure for the insoluble volatile fatty acids obtained, then :

(A-16) × 2 = percentage of cocoanut-oil in fat taken.

The process is rapid. From beginning to end it requires about two and a half hours, and during most of that time very little personal attention is called for. The manipulations involved are few in number, and are not complicated or tedious, and the apparatus employed is simple and readily set up.

The authors have under consideration the application of the process to the estimation of cocoanut-oil in admixture with margarine.

Inks and Forgery. In the October number of Knowledge Mr. Ainsworth Mitchell, who it will be remembered recently gave evidence in a case dealing with a forged will, has a striking article on inks. Incidentally he lets his readers into the secret as to the tests which he applies in order to discover the age of any particular piece of writing. If it is shown that the ink on a document purporting to be drawn up, say, ten years ago, is really quite fresh, then there is every chance that the writing has been forged. An examination of the colour of the ink may be the sixth day, and in after years its age may be told when helpful. Blueblack ink may be recognised as fresh up to the blue provisional pigment has faded and left only the black. The blue colouring as time goes on is hardly acted upon by reagents, but for a year or two it is. In fact, writing done within that time will at once diffuse if treated with a 50 per cent solution of acetic acid, whereas when it is five or six years' old, diffusion, if it takes place at all, is very slow and limited in extent. A still more useful reagent is a saturated solution of oxalic acid, which causes the pigment of relatively fresh writing to give an immediate smudge, but has very little, if any, effect on writing six or eight years old,

escape from the top of the condenser tube. Water is now admitted to the disc-heater by opening the cock, T1, and, supposing it is desired to take a refractometer reading at 40° C., carefully adjusting until the flow of water is sufficiently retarded to indicate a temperature of about 45° C. at the refractometer thermometer. Directly the thermometer is steady, cold water is admitted by the cock T2 -the rate of its entry being at once seen in the glass mixing-chamber. This flow is adjusted until the thermometer indicates 40° C., when the latter will remain at this temperature for as long a period as the apparatus is kept working.

If after this first determination it be now desired to refract an oil, say, at 25° C., this temperature may be quickly attained by increasing the flow through T2, and again carefully adjusting until the refractometer thermometer indicates the desired point.

The writer has had the apparatus in use in his laboratory for the past month, and found it to answer all requirements perfectly.

RESEARCHES UPON THE COMPLEXITY
OF TELLURIUM*
By WILLIAM R. FLINT.

Introductory.

IN two papers from the Kent Chemical Laboratory of Yale University a description has been given of a hydrolytic method for the fractionation of tellurium (Browning and Flint, Am. Journ. Sci., xxviii., 112 and 347). An account also has been given, in the second paper, of results obtained from a preliminary application of the process, in which it was indicated not only that the substance might really be complex in spite of the almost conclusively negative evidence hitherto adduced, but also that its anomalous position in the periodic system might be due to the actual presence, as Mendeleeff predicted, of a constituent having a higher atomic weight.

In the work referred to, the fractionation was so conducted as to produce both end fractions simultaneously, and since the method employed was practically that of fractional crystallisation, this involved, unless the operation had been long enough continued, the introduction into each of these portions of material from the middle fractions.

It was therefore determined to modify the fractioning process in such a way as to secure a more rapid separation of each end fraction independently, and to apply it to a much larger amount of material.

PART I.

As a check upon the whole work 150 grms. of the purified product were dissolved in nitric acid (sp. gr. 125) and crystallised out as basic nitrate by evaporation of the solution at about 85° C. After being washed with concentrated nitric acid, the basic nitrate was dried by heating at 110° to 120° C. in specially constructed electric drying ovens, through which was passed a current of air dried by sulphuric acid. The air was bubbled first through a solution of potassium hydroxide, then through concentrated sulphuric acid, and finally filtered by passing through a tower containing cotton-wool.

Portions of this preparation were then weighed out in platinum and brought to constant weight by heating, for periods of about twelve hours each, at 140° C. in air dried as just stated, in the electric heaters. The temperature was next raised, and the nitric acid expelled from the compound by ignition up to 425° to 450° for twenty to twentyfour hours. The experiments were finally completed by a short ignition over a Bunsen burner at such a temperature as would ensure a glassy condition of the fused dioxide.

In all the basic nitrate experiments detailed in this paper, the weighings were made by the method of oscillations, a standardised set of weights employed, and a correction for the inequality of the balance arms applied. Lastly, the weights were reduced to a vacuum. The corrected results of the determinations made upon the unfractionated material, which, excepting No. 5, were performed in pairs, are given in Table I. It should be added that in no case was there evidence of the escape of traces of dioxide by volatilisation.

It may be noted that the mean atomic weight, 127:45, is practically identical with that found by Norris (Journ. Am. Chem. Soc., xxviii., 1675), and that the concordance of these experiments is quite as good as those given by him. The close agreement of this result with both the figure given by Norris and also those found by numerous other investigators indicates a complete correspondence in quality between the material employed in the present research and that used by others. It also confirms the general conclusion that the atomic weight of tellurium, when prepared by the ordinary methods, is 127.5.

The Fractionation by Water Hydrolysis, and Atomic Weights of the Fractions Obtained.

(See accompanying plan).

Five hundred grms. more of the re-distilled tellurium were oxidised by nitric acid and converted to tetrachloride by repeated evaporation with hydrochloric acid. The solu tion, containing as little excess of acid as possible, was, in four portions, diluted with about 4 litres each of boiling distilled water, and allowed to cool. The white crystalline dioxide which separated out was collected on a filter and washed. After solution in hydrochloric acid (in slight excess) it was re-precipitated by abundant dilution as before. When the fourth repetition of this process had been made, using in each case only the dioxide which had been previously precipitated by dilution, a small portion of the fourth precipitate was taken out as a sample. This was crystallised from nitric acid by the usual method, and a second series of basic nitrate experiments, conducted in every way like those of Table I., was performed. The data of these are found in Table II.

10.23038 Mean 125°37

Fraction ten consisted of but 23 grms. of dioxide. Since the fractioning process had reduced the quantity of dioxide from over 600 grms. to a little more than 20, it seemed possible that the limit might have been reached. About 15 grms. were consequently transformed to basic nitrate, in two separate preparations. The purpose of this was to eliminate, if possible, the objection that this lowering of the atomic weight might be due to lack of constancy of composition of the basic nitrate. The first sample, the data of whose analysis by the basic nitrate method (applied exactly as described for the preceding preparations) are given in experiments 1 and 2 of Table IV., contained about 5 grms.; the second, experiments 3 to 7, 16 grms. TABLE IV.-After Ten Fractionations. 2TeO2. HNO3 taken.

2TeO2 found.

Exp.

Grms.

Grms.

Atomic weight of tellurium.

of Table I.

A comparison of the mean values given in the four preceding tables shows a remarkable lowering of the atomic weight with the progress of the fractionation. In considering the difference between the mean result, 1243, of Table IV., and that, 125'4, of Table III., it should be noted that, since the material which gave the latter figure was derived from the filtrate of fraction eight, it was to have been expected that its atomic weight would be higher than that of fraction eight itself. Owing to the small quantity remaining of fraction ten, it was impossible to make another direct fractionation. But the dioxide produced in the experiments of Table IV. was dissolved in hydrochloric acid and the hydrolysis repeated once more,

HCl + H2O

TeO2 HNO3

The

the precipitate being converted to basic nitrate. basic nitrate experiments gave 124.62 and 124.64, but these figures cannot be accepted as conclusive. In the first place, they were made upon material which had already been used for previous determinations, in which the dioxide formed had been heated in platinum up to fusion. During the course of the basic nitrate experiments described in this paper, a progressive, but minute, loss of weight was observed in the platinum crucibles used. This fact suggests the possibility that the tellurium may have become contaminated with a little platinum, which would of course raise the atomic weight found in these two experiments. And second, while these two experiments were in progress, an extreme humidity unfortunately characterised the weather. The tendency of this condition would be to increase the atomic weight also, since traces of moisture absorbed by the basic nitrate during those short, but necessary, periods in which it was out of the heaters and desiccator for the purpose of weighing, would tend to