SOME ASPECTS OF THE ANALYSIS OF HOUSEHOLD DETERGENTS 35•) phosphate is converted to the diacid form and precipitated with silver ions, it yields two hydrogen ions per molecule: H2PnO3n+•n--k (nq-2)Ag+•Agn+2P•O3n+1 • +2H +. If another sample is first hydrolysed and then treated similarly, it yields two hydrogen ions per phosphorus atom, or 2n hydrogen ions per molecule: Na•+•P•Oa•+•--•n NaH2PO4--•n AgaPO4 • +2nil +. If the two experiments are done on equal aliquots of the same solution, and the liberated acid is titrated, the ratio of the two titrations is equal to the mean chain length. The procedure is as follows: A sample of the detergent powder is rendered free from surface active matter by repeated extraction with boiling alcohol. After being dried, a sample containing about 1 g of P•O5 is dissolved in water and diluted to 250 mi. A 25 ml aliquot is boiled under reflux with 25 ml of 3 N nitric acid for at least 30 minutes. Meanwhile a second 25 ml aliquot is placed in a beaker and made slightly acid with dilute nitric acid. pH electrodes are inserted, the solution is slowly titrated with 0.1 N alkali, and the pH is plotted against volume. When a complete step has been plotted, the pH at the steepest point is determined, and the pH of the solution adjusted to the value found. The condensed phosphate is then present as the diacid salt, NanH•P•Oan+•. 10 ml of 10% silver nitrate solution is added, and the liberated acid is titrated to pH 6.0 with 0.1 N alkali. Let this titration be V• mi. The refluxing solution is then cooled and its pH is adjusted to 4.5, so that the phosphate is present as NaH•PO4. Again 10 ml of 10% silver nitrate solution is added, and the liberated acid titrated to pH 6.0 with 0.1 N alkali. Let this titration be V• mi. Then the mean chain length = V•/V•. It should be noted that although the two aliquots of the phosphate solution must be exactly equal, it is not necessary to know the exact weight of material used or the exact normality of the alkali. This method has been tested on a number of pure phosphates and on a phosphate recovered from a detergent and analysed by chromatography. The values obtained were: Tetrasodium pyrophosphate pentasodium triphosphate hexaguanidinium tetraphosphate recovered phosphate do. by chromatography 2.00 P atoms/mol 3.02 P atoms/mol 4.00 P atoms/mol 2.66 P atoms/mol 2.67 P atoms/mol If a calomel electrode is used as the reference electrode for this type of titration, it is desirable that it should be of the sleeve type, as the wick

360 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS and sintered disc types tend to become clogged with the precipitated sil,rer chloride and consequently give false readings. APPLICATIONS OF FLAME PHOTOMETRY 1. Sulphate The familiar gravimetric method for sulphate can be applied to deter- gents, but it is very tedious because both surface active matter and silica must be removed before the barium sulphate is precipitated. A rapid method of relatively low precision has been developed •ø'n in which barium sulphate is precipitated, dissolved in EDTA and determined by flame photo- metry, using the E.E.L. Flame Photometer. This is a low-sensitivity in- strument which uses filters to select wavelength, and a low flame temperature to minimise interferences. This means that it is not really suitable for the determination of metals other than the alkalis and calcium, which are easily stimulated and give strong radiations. It has, however, been found possible to determine barium, using much higher concentrations than usual (several hundred parts per million as against the usual ten or twenty). A sample of the detergent containing 0.25--0.50 g of sulphate as SO• • is dissolved in water. Excess hydrochloric acid is added and the surface active substances are extracted with ether. The aqueous layer is diluted to 250 ml and mixed. A 5 ml aliquot is pipetted into a 10 ml centrifuge tube, and a 5 ml aliquot of the standard (sulphuric acid solution containing about 0.2% or 2000 ppm of SO• • and accurately standardised) into another. 5 ml of 1% barium chloride solution is added to each tube. After mixing, the precipitates are centrifuged out and washed twice with water to remove sodium ions, which cause serious interference. They are then dissolved in 1 •o E.D.T.A. solution made alkaline with ammonia, and diluted to exactly 10 ml still in the tubes, with the same reagent. The filter used in the flame photometer is a Wratten 65A, which transmits the barium line at 493 m•. The ammoniacal E.D.T.A. is used to set the zero, and the treated standard to set the galvometer to full scale deflection. The test solution is then read. The concentration of sulphate ion in the detergent is given by scale reading x standardisation factor on H2SO• % SO,F = 2 x weight of sample The calibration curve for sulphate is linear to at least 1000 ppm. The precision is about 4-2% of the concentration found for a scale reading of 50, or 4- 0.4}/0 if the sodium sulphate content is 20%. This is poor compared with the gravimetric method, but a complete analysis takes only about 45 minutes instead of several hours.