202 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS prevent fouling of the main column. The resolution for the solvents and pro- pellants of hairspray aerosols was satisfactory, except for Propellants 12/114, and for isobutane/dimethylether (Table IX). Before proceeding with the analysis, some important questions--specific to this type of analysis--had to be considered. How reliable is a glc-injection with the system? To answer this question one transfer vessel was filled with a hairspray aerosol sample (+ internal standard) and 10 successive injections of 20 I•1 were made. From the integrated glc-data the •o of the solvents and propellants were calculated. Statistical evaluation gave the following coefficients of variation: P12 (1.4•o), P11 (1.3}/o), DCM (1.5•), EtOH (2.5•). It can be concluded that the sampling and injecting system is quite reliable (Table X). Table X. Reliability of 10 successive injections of the same liquid mixture One small (10 ml) all-glass aerosol bottle was filled with a hairspray aerosol sample and an internal standard added. The liquid mixture was analysed by 10 successive injections according to the method and the integrated peak areas were used to calculate the % of the volatile com- ponents. The calculated % are: 1 27.4% P12 54.8% Pll 5.25% Dichloromethane 6.76% Ethanol 2 27.1 54.5 5.16 6.79 3 27.3 55.0 5.22 6.91 4 27.5 55.2 5.45 7'21 5 27.2 55.5 5.25 7.14 6 26.8 54.1 6.34+ 6.74 7 26.5 53.5 5.76+ 6.69 8 27.3 55.1 5.23 7.10 9 27.8 55.9 5.27 7.02 10 27.3 55.0 5.24 6.92 Mean 27.2 54.9 5.26 2.69 Standard deviation 0.4 0.7 0.08 0.20 Variation coefficient 1.4% 1.3% 1.5% 2.5% + Abnormal results. Experimental Can an aerosol sample be analysed at any emptying stage, or in other words will the composition of an aerosol change during use? To find the practical answer one container was sampled several times according to the method, while releasing it between the sampling. The results of the analysis of the five small aerosol bottles for the intermediate transfer are given in Table XI. It proves that sampling can be done at any emptying stage, except when the can is nearly empty.

RISK-BEARING SUBSTANCES IN COSMETICS 203 Table XI. Composition of a hairspray aerosol during its use A single hairspray aerosol is sampled five times for analysis, at different stages of emptying. The following data were obtained: 'Full' 'Empty' = 530 g Removed Removed Removed 30 g net weight 200 g 300 g 400 g left Propellant 12 29.8% 29.4% 29.3% 28.5% 23.8% Propellant 11 45.0 45.1 45.1 45.8 46.6 Acetone 0.30 0.29 0.30 0.23 0.28 Dichloromethane 22.7 22.6 22.7 22.5 26.6 Ethanol 4.4 4.4 4.0 4.1 4.7 Is the injected volume of importance? Generally speaking, the introduction of an internal standard makes the volume of the injected liquid of no importance. But in this kind of analysis another important factor must be considered, namely that the 'dead' volume of the needle must be negligible in comparison with the injected volume. Experimentally the minimum volume was 15 pl. In all our experiments the volume is standardized to 20 I•1 to obtain optimal reliability. Is one general standard (cyclohexane) acceptable for such a mixture of dif- ferent kinds of organic compounds (alkanols, alkanones, hydrocarbons, fluorin- ated and chlorinated hydrocarbons)? The linearity of the response for certain practical concentration levels was checked by analysing three different reference Table XII. Reliability of the use of cyclohexane as a general internal standard Three different standard mixtures, containing nine compounds and cyclohexane were analysed by glc. From the integrated peak areas the correction factors F were calculated by the formula F = xdS) (As/A0. % by weight Correction factors Concen- tration A B C A B C levels Propellant 12 39.3 21.4 28.4 1.52 1.53 1.47 20-40% Propellant l 1 17.0 8.9 26.6 1.58 1.55 1.53 10-30% Acetone 6.7 18.0 13.3 0.86 0.84 0.80 5-20% Methanol 1.7 4.6 3.2 0.77 0.74 0.72 1-5% Dichloromethane 13.2 18.0 6.6 1.19 1.12 1.10 5-20 % Cyclohexane 6.7 6.1 6.6 I 1 1 Ethanol 3.3 4.5 1.9 0.78 0.81 0.79 2-5% Methylchloroform 6.8 12.4 9.9 1.55 1.57 1.57 5-15% isoPropanol 3.3 4.6 1.9 0.84 0.87 0.84 1-5% Nitromethane 2.0 1.5 1.6 1.03 1.00 1.18 1-2% Conclusion: For these levels of concentration cyclohexane as a general internal standard is acceptable, except for nitromethane.