190 H.R. Watson et al. Just as bmenthol has the greatest cooling activity of the eight stereoisomeric menthols, we believe that carboxamides such as II (Fig. 2) with the same stereochemistry also possess higher activity than any of their isomers. Thus, d-menthol is 45 times less active than its natural enantiomer/-menthol (as determined by the oral threshold method), and the d-enantiomer of Compound II is 70 times less active than II. These observations are consistent with menthol and the synthetic coolants acting at common receptor sites, the sites being associated with the nerves responsible for the sensation of cold. We believe that whare stereoisomerism is possible in a molecule it is a determinant of activity. GENERAL PROPERTIES OF COOLING COMPOUNDS Apart from the cooling effect, and a degree of flavour potentiation and odour modi- fication,* there is no common property of cooling compounds. For instance, there is no association between minty smell and cooling. There being such a wide range of chemical type, the physical properties of the compounds are varied. Most are solids, although liquids are not uncommon when the functional group is hydroxyl, N,N-dialkylcarboxamide, or phosphine oxide. Most are readily soluble in common organic solvents, but owing to the requirements of the log P value, all are of very limited solubility in water (although it should be noted that saturated aqueous solutions have readily perceptible cooling effects). The alcohols and N,N- dialkylcarboxamides are in general volatile, and are odorous, the odour types ranging from minty to fruity to earthy to camphoraceous. Compounds of other classes have low or very low volatility and are non-odorous. Many of the N-alkylcarboxamides, phosphine oxides and ureas are virtually involatile, and are odoudess and tasteless. Chemical and thermodynamic stability follow the known characteristics of the constituent groups. The simple alcohols, acids, amides and phosphine oxides are of very high stability. This includes the long-term stability in aqueous environments which is essential for most projected end uses. Sulphur compounds and ureas are less stable and careful consideration is needed when matching these compounds with usage compositions. HUMAN SUBJECTS AND THE COOLING EFFECT TESTING OF RELATIVE ACTIVITY Intrinsic Activity- Oral Thresholds A reliable test of the intrinsic cooling effect of a compound was essential for development of any theory of relationship of structure to activity. The tongue is very sensitive to the cooling effect, it offers a convenient test site and, in particular, appears to present re- latively little physical barrier to the compounds. Although there are a few exceptions, cooling compounds applied to the tongue appear to penetrate rapidly to the cold nerve * It appears that when cold receptors are chemically excited, there is an associated mild response from adjacent odour and taste receptors, and modification of these two senses is almost certainly a general characteristic of cooling compounds (25). (David Kendall of Arthur D. Little, Boston, has tested over 100 compounds of our series all modified flavour.) The property is in accord with that of pungent or 'warming' compounds pungent components of peppers, chillies etc. are well established as flavour modifiers.

New compounds with menthol cooling effects 191 receptors. For this reason, we believe that the oral activities are a reasonably close approximation to the intrinsic activities of the compounds. Oral equipotency tests, in which trained panelists would derive equivalent cooling effects by balancing different concentrations of test and control solutions would have been the most scientifically satisfying, but such tests are very time-consuming and in view of the large number of compounds involved, were not considered for our own studies. (Trained panelists at Arthur D. Little, Boston, derived the same order of ranking of effectiveness which results from the threshold method described below when testing 20 of the compounds by an equipotency method.) A threshold method was used for all determination of oral activity, and the values of oral threshold derived from it are treated as (reciprocal) measures of intrinsic cooling activity. Filter paper (1 x 1 cm) was impregnated with a known amount of compound by application of a measured volume (1-5 pl) of a solution in redistilled petroleum ether (40-60ø). After 30 sec, the paper was placed on the tongue of the subject, who was required only to report presence or absence of a cooling effect. After an interval, the procedure was repeated, adjusting the amount of compound as required, until the threshold was ascertained. Usually, a sufficiently accurate estimate could be obtained with as few as five to ten measurements, and although there was a variation of + 30•o on repeat tests, this was acceptable because differences between compounds were usually quite considerable. Individuals differ in their sensitivity to the cooling effect, and the personal thresholds of a group of subjects are likely to cover a considerable range. For example, the menthol thresholds of 23 subjects, chosen at random, were in the range of 0.02-10 lsg. Also it is a general feature that the distribution is not simple Gaussian. Most panelists had thresholds reasonably near the minimum, there being a long tail to the value representing lowest sensitivity. Even though the individual members of a panel may have had a wide variation in their menthol threshold, they ranked a series of compounds in the same order. Thus, although the basic sensitivity to a cooling effect varied from subject to subject, the relative effects of a series of compounds were equivalent for all panelists. All strong compounds appeared relatively strong to all subjects, and all weak compounds appeared relatively weak. In the absence of this feature, ranking of intrinsic effect would have been much more difficult. In view of the wide variation between panelists, arithmetic means of threshold results were obviously of no value. Geometric means gave useful ranking, but it is believed that a more accurate comparison of compounds was achieved by comparing a panelist's threshold for a test compound with his personal threshold for a standard compound. The average ratio for the whole panel was then derived as the arithmetic mean of the individual ratios, and the mean threshold was calculated by reference to the known mean threshold (geometric mean of 23 subjects) of the standard substance. Menthol was not chosen as the standard substance because it gives occasional inaccurate values owing to its volatility N-ethyl-p-menthane-3-carboxamide (II, Fig. 2) is our standard. The above method appears to be reliable. Panelists were routinely checked for standard response by normal cross-check methods, and it is noteworthy that their thres- hold for a given cooling compound does not change even after several years of practice. Oral thresholds recorded during the work spanned the range 0-08-50 pg. A compound with a threshold of less than 0.5 [tg is regarded as 'strong'. Thresholds of greater than 50 [tg were not quantified at that level a compound shows exceedingly weak activity, and