280 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS THE LECTURER: Variations between instruments of the same type will be reason- ably small but between instruments of different types the variation between the readings is fairly high. This is because instruments differ in measuring techniques, i.e., they view the patterns in different ways. Some look at the pattern in a given direction, others view the pattern over all directions as in the integrating sphere type of instrument. Differences between the absolute readings on different instruments are not important when colour differences are being measured on a particular instrument. Provided the reproducibility of measurement is good colour difference measurements will be reliable. I do not have any information in this respect for the Spectronic 20. It may also be used as a spectrophotometer but has the disadvantage that it measures with a fairly broad waveband, i.e. 20 nm, which may give rise to errors when measuring reflectance curves with sharp peaks. DR. A. W. MIDDLETON: Could you give us the reference to the Simon and Goodwin paper, and could you tell me how you sort out metameric colours since I assume that one of these instruments measures all its colours in one reference light? THE LECTURER: Generally speaking all the instruments I have referred to will measure colour under standard illuminant 'C' which is designed to correspond closely with daylight. Some ooe the instruments are also capable of measurements under other illuminants, i.e. tungsten lighting. With these instruments it is possible to assess colour differences that would be apparent under both daylight conditions and tungsten, and hence the degree of metamerism. The accuracy of assessment of metameric patterns is somewhat limited by colorimetric means. It is also a difficult problem with visual judgments because when metameric pairs of patterns are compared, the differences between observers' visual characteristics begin to show up very markedly, and widely differing opinions about a colour match may be obtained particularly when there are large age differences between observers. I certainly do not think instrumentation is the answer to the problem because instruments are only designed to have a response approximately equal to the visual response of the average observer. It is a reliable proposition only when comparing patterns which are non-metameric, for example where the standard is dyed, or coloured, with the same colorants as the sample. In practice a standard is set up from production at some stage, and compared instrumentally with subsequent production material. The reference you require is: The rapid graphical computation of small color differences by F. T. Simon and W. J. Goodwin, published by the Union Carbide Corporation, U.S.A. MR. J. M. TRmCLE: I have carried out an investigation of several differential colorimeters with a view to using theIn for quality control in industry, and there is a large spectrum of colours in eye cosmetics and face powders which are very low in green reflectance and the G value is well down. Even the most sophisticated colorimeters that I have looked at, do not seem to be able to perceive the differ- ences which must constitute say, two or three MacAdam units, where there is a G value lower than say 20%. Is this a fact of all differential colorimeters? Are there more sophisticated ones than, say, the Colormaster which I have looked at and which failed miserably in this area?

INSTRUMENTAL COLOUR MEASUREMENT AND CONTROL 281 THE LECTURER: I am rather surprised that the Colormaster is not able to fairly accurately estimate colour differences on patterns with the sort of reflectance you mentioned, i.e. 20%. One has to accept that for very dark colours, •vhen the reflectance is very low, the sensitivity of the instrument begins to be a slight problem, but I do not consider them any more significant than the problems that occur in visual assess- ment at low reflectance. I can not really offer an explanation for your observations. As far as we could discover there have not been any limitations for any particular type of colour with the instruments that I have mentioned, including the Colormaster. MR. j. M. TRIGGLE: Did you say that the efficiency of the red and blue response is in any way inter-dependent upon the green response in the Colormaster? In other words, that if the efficiency of the green will go down, the efficiency of the red and blue will go down as well? THE LECTURER: There is no inter-dependence of the red, green and blue responses. The responses of the Colormaster are designed to be approximately equal to that of the eye. And this is so, of course, with other instruments, the colour eye and the Zeiss Elrepho. It is true to say that the Zeiss Elrepho and the Color-Eye do approximate better to the average eye characteristics but I do not think that this would make any great difference when you are comparing non-metameric colours. MR. j. M. TRIGGLE: May I just substantiate what I said. This work was carried out in conjunction with the Paint Research Station. and we intentionally adulterated some mascaras and eye cosmetics by adding red pigment the Colormaster did not change its reading with the red filter at all, although the colour change was visually quite perceptible. THE LECTURER: I find this quite surprising, but it may be worth looking more carefully at the method of presentation of the samples to the Colormaster which may be giving rise to errors of reading large enough to mask the small colour differences. DR. j. J. MAUSNER: It may well be that the difficulties described by Mr. Triggle are due to the fact that a colour instrument which measures three definite wave- lengths in a spectrum is just not sensitive enough, and one can well envisage a theor- etical case of a colour where the reflectance at three different wavelengths are equal to standard, but where the rest of the colour spectrum is different. THE LECTURER: This sort of measurement is only suitable for non-metameric matches. If you are quite sure that the colorants in the standard colour and the pro- duction are the same, then you should be fairly safe in looking at measurements at, say, three wavelengths and comparing these, since in an identical match the reflect- ances at tlm-ee wavelengths should be the same. The Colormaster, however, does not have these defined wavebands of measurement in fact it looks at colour with a response which extends right over the spectrum. True that you have a green, a red and a blue filter and they have predominantly sensitivities in those regions of the spectrum, but nevertheless they do extend over a broad region.