ERRORS IN THE ANALYSING OF PERFUMERY RAW MATERIALS the sum of the cross-sections of their constituent atoms. This is claimed to be true not only with hydrocarbons but also with compounds involving O, N, Br, etc. Nevertheless, the errors between calculated and real values become significant for types of substances frequently found in perfumery raw materials. Each type possesses broadly a corresponding proportion- ality factor K. In effect one has T (total ionisation) = K (n c Q• q- n• Q• q- np Qip etc.) where n = the number of atoms of carbon, of hydrogen, and of another atom p, and where qi represents the ionisation cross-section of the atom in question. Once again we are faced with the necessity of calibrating by means of internal standards when using these detectors. A frequent problem in the analysis of perfumery raw materials is the evaluation of minor constituents. An obvious approach is to increase the sensitivity of the detectors. With ordinary equipment, the quantity nor- mally detectable is of the order of a microgram, whether a catharometer or an ionisation cell is used. The noise level with 25 m.c. - øøSr sources amounts to approximately 0-5% of full scale deflection, so that it is about the same as that of the most sensitive type of catharometer. Noise is, however, some- what more apparent on the E-ray charts, as it has a shorter period than in the case of a catharometer. One could think of amplifying the signal by augmenting the quantity of product injected, but in this event and even with a detector of adequate sensitivity, the trace constituent is frequently eluted with the very enlarged and diffused band of a major constituent, and becomes very difficult to measure. One can increase the efficiency of the separation by increasing the length of the column, but this also increases the size of the peaks. One is therefore left with the necessity to concentrate the impurity by preliminary gas chromatography or by a chemical pre- treatment, but this raises a number of new problems. The error most frequently committed in the absence of internal standards is the assumption that the product utilised has entirely vaporised. In fact, perfumery raw materials such as essential oils are only partially vaporised, because they either contain heavy natural constituents or undergo changes. That is why analyses having no regard to these considerations give, for example, excessive values for lemongrass and citronella oils. In almost all cases, the use of an internal standard becomes necessary. One should not by any means assume that all the eluted substances must be present in the same proportion, if at all, in the injected sample. In effect, under the conditions of chromatography, a certain number of original con- stituents can change in contact with the equipment, with the column packing •or by reaction amongst themselves.

36 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS The construction material of the columns must be considered in par- ticular. Stainless steel is used preferentially, or, in our opinion, it is even better to use glass. 3/Ionel metal and copper can give rise to variations, notably when the product being analysed is phenolic. The choice of materials for the injector and for the detector must also be considered. At high temperatures, metals can promote catalytic effects such as cracking or dissociation. The effect of the charges (supports) has been most frequently men- tioned TM. The use of alumina and of silica gel as the support is rarely indicated in the treatment of perfumery raw materials. Diatomaceous earth (Celite) and firebrick are used most often. Specially treated quartz has recently been used with advantage by Japanese workers. fi-pinene is isomerised into a mixture of a-pinene, camphene and limonene on silicone supported by firebrick or Celite. Aging of the columns reduces the isomerisation on firebrick and suppresses it on Celite. The replacement of silicone by a polyethylene glycol prevents isomerisation 29. According to others, both a- and fi-pinenes react when firebrick or Celit• are employed, whereas Chromosorb is inert TM. It has been stated that isomerisation and partial destruction of terpenes take place above 160 ø when silicone on Celite is employed and even above 140 ø when firebrick (Sterchamol) is the support a•. Nerol is dehydrated towards 190 ø on Apiezone L supported by acid washed Celite, with the formation of myrcene and ocimene a2. For the dehydration of alcohols on charges supported by firebrick and Celite see Liberti et al aa. On a strongly polar column fi-citronellal is cyclised into isopu!egols to such an extent that the analysis of mixtures of a- and fi-citronellals with isopulegols becomes very complex a4. The mixed isopulegols and a- and •- citronellals are determined by means of a neutral Chromosorb W column, a-citronellal is estimated with an acidic Celatom F.M.73 column, which transforms fi-citronel!al into isopulego!s, and the fi-citronellal is calculated by difference. Octene-l-ol-3 partially isomerises at 125 ø into ethylamylketone a•. Ethylene isomerisations which take place when firebrick is employed are almost negligible where acid-washed Celite (Embacel) is used a6. Citrals can be transformed by certain supports into p-cymene, and terpineol into dipentene a*. Peroxides, which occur frequently in perfumery raw materials as the result of chemical changes, are degraded with the formation of carbonyl products, acids, etc. The polymerisation of monomers can be retarded by the incorporation