236 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS The carbonium ion, or biological equivalent, from the cyclisation may undergo many alternative processes depending on the plant or conditions. Simple hydration gives a-terpineol (19, 20) or an elimination of a proton may give A4-carene, limonene (21-24), or terpinolene. If the remaining double bond is involved in cyclisation the nexv "ion" may give borneol, a-pinene (19, 23-27), or [I-pinene (10, 23, 24, 27) or rearrange further to give camphene or fenchol. Alternatively a rearrangement of the "carbonium ion" by a 1,2-hydride shift gives a system which could give sabinene (19, 24, 28), or sabinene hydrate (24). Thus seven skeletal types may be genera- ted from the one "carbonium ion" by elimination of a proton or hydration with or without rearrangement. However, of course, individual plants may only use one or a few of these processes. Experimental work on the biosynthesis of cyclic monoterpenoids (Fig. 5) is limited. Present evidence suggests (19, 27) that many of the reactions, including some of the further oxidation processes, may be reversible. The further metabolism of these basic skeletal types may involve a combination of processes such as reduction and oxidation to give the plethora of mono- terpenoid structures. For example a-terpineol may be considered (Fig. 6) to cyclise to 1,8-cineol. However, although there is good evidence for the terpenoid origin of 1,8-cineol with the expected labelling from the incorpora- tion of [1-14C]geranyl pyrophosphate (29), when Arigoni (30) studied the utilisation of [2-14C]mevalonic acid he found complete randomisation. Presumably the precursor had been degraded to C02 by the plant before the precursor could reach the site of the relevant enzymes. •l• c• HO 0 o OH 1,8-cineol a sabir•al thujone t hymol Figure 6 Another problem was encountered by Banthorpe (28, 31) when he studied camphor, sabinol and thujone biosynthesis (Fig. 6). Apart from a small amount of randomisation he found that up to 87% of the label •vas at one of the two positions instead of 50%. This result may be interpreted by postulating a pool of unlabelled dimethylallyl pyrophosphate in the plant

THE BIOGENESIS OF TERPENOID ESSENTIAL OILS 237 which dilutes the precursor so that in the biosynthesis of geranyl pyrophos- phate (Figs. I and 2) from [2-14C] mevalonic acid C-4 will be more radioactive than C-8. A similar reason may explain the results of Sandermann (21, 32) when he studied limonene (Fig. 5) and pulegone biosynthesis (Fig. 7), and possibly his results with a-pinene (26) and thujone (33). In the other cases • • '-o -•--o menthofuran pulegone menthol 0 i 5o merithone Figure 7 where the labelling pattern has been examined [carvone (22) and thymol (34)] the two positions are not clearly distinguished. In the essential oil field probably the most studied system is Mentha piperita. Unfortunately the experimental problem necessitates much of the work being conducted with 14CO 2. However, this precursor does allow the study of the variation in concentration, specific activity, and total activity, with time (24, 35-37). Their conclusions are summarised in Fig. 7. A limited amount of work supports some of the steps, with the incorporation of piperitenone (37), pulegone (37, 38), and menrhone (36). The general scheme seems to involve allylic oxidation to piperitenone, presumably via piperitenol, followed by successive reduction of the double bonds and finally the ketone function (or oxidation of pulegone to menthofuran). Several other monoterpenoid skeleta are encountered in essential oils.