122 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS troscopy, Edwards et al. (12) confirmed the earlier observation of Dent (9) that light penetration into whole skin was proportional to wavelength. Jacquez and Kuppenheim (13) confirmed that the erythemal condition was associated with the preferential ab- sorption from a white light source of green light by hemoglobin. However, the demon- stration by Jacquez et al. (14) that electronic measurements of intensity could be made to an absolute accuracy of 1.0%, clearly showed the potential of these devices to mea- sure very small changes in reflected light intensity. In the last two decades a number of small portable devices have been used to measure erythemal values in skin (15,16) with an accuracy and discrimination superior to that of the human eye (17, 18). The remaining problem when considering the evaluation of the clinical condition of skin is the impartiality of the electronic measuring device. Unlike the trained observer, a photosensor has no discretionary powers. Light from the surface of the skin containing topographical information cannot be measured to the total exclusion of light reflected from the bulk tissue containing information on physiological response. It was the aim of this investigation to construct an illumination system for the assessment of the clinical condition of skin which would highlight the surface condition to the exclusion of the erythemal condition or vice versa. To this end a polarized illumination system was developed. EXPERIMENTAL POLARIZED PHOTOGRAPHY Although preliminary photography was successfully developed using a Polaroid SX70 camera to produce instant color prints, it was found that better results were obtained with higher quality photographic equipment as below. The skin area was illuminated from opposite sides at approximately 20 degrees by two Xenon flashlamps (Broncolor Impact) and photographed from 1 m with a Has- selblad camera and 120-mm telephoto lens at f32 using Kodak Vericolor-2 type-S film. At least three photographs were taken of each specimen. A standard photograph for reference purposes was taken. Photographs of reflected polarized and depolarized light from the same specimen illuminated by polarized light were also taken. To achieve this, the system was modified by placing, over each flash lamp, a polaroid filter (18 inches by 18 inches) orientated to make parallel both the polarization vectors of each light beam and also the horizontal specimen plane. A third, rotatable, polaroid filter was fitted to the camera lens. When set parallel to the other polaroid filters, a polarized reflected light image was recorded whilst a perpendicular orientation gave a depolarized reflected light image. In each photographic field a bright aluminium plate partly covered with Kodak white reflectance coating was included as a reference for polarized and depolar- ized light intensity. REFLECTION SPECTROSCOPY Diffuse reflectance spectroscopy was undertaken with a Perkin-Elmer Lambda 7 UV/ VIS spectrometer, at a fixed 4-nm bandpass interfaced to a Perkin-Elmer 3600 UV data station. This standard transmission spectrometer was modified with an external inte-

OPTICAL DISCRIMINATION OF SKIN 123 grating sphere attachment with quartz fibre optics. The attachment was fitted to a custom-made support frame (Neo-Tech Engineering) to ensure a light-tight contact of the sample port of the sphere onto the skin of human volunteers. Polarized reflection spectroscopy was undertaken with a polaroid filter fitted over the sample port, thus allowing only polarized reflected light, from the skin, to re-enter the integrating sphere. With this arrangement it was not possible to measure the spectrum of the perpendicularly polarized reflected light and an alternative procedure was used. In vivo polarized reflection spectroscopy of human skin was undertaken with a modified Jobin-Yvon JY3CS computer-controlled spectrofluorimeter. An incident illumination assembly for use with thin-layer chromatography plates replaced the conventional cell compartment. An off-axis concave mirror was placed as close as possible to the normal axis of the incoming incident light beam to collect reflected light efficiently. In this way very small changes in the sample-to-mirror distance (5mm) did not critically affect the measured intensity ([2.5%]), as was the case with the design originally supplied ([10%]). Visible-grade polaroid filters were mounted immediately after the exit slit of the excitation monochromator and before the entrance slit of the emission monochromator. Reflection spectra from sites on the palmar surface of the hand were recorded by running both monochromators synchronously over the same spectral range. Four nanometer slits were used throughout, and in this way spectral bandpass compati- bility was maintained with the Perkin-Elmer Lambda 7 spectrophotometer. Parallel and perpendicular polarized reflection spectra were recorded and are referred to by the orientation vectors of the polaroid filters in the incident and reflected light beams. Since this was a single-beam system, all reflection spectra were corrected for the wavelength-dependent transmission of the polaroid filters by recording the spectrum of a diffuse white standard. This standard was prepared by painting a thick film of Kodak white reflectance coating onto a bright aluminium plate 7 cm by 7 cm. The recorded, paired spectra of the skin were then normalised by setting the parallel polarized reflec- tion value at 700 nm to unity. In this way only relative spectral intensity changes could be compared. TRANSMISSION SPECTROSCOPY The methods used to measure the spectral transmission characteristics of polaroid filters were determined by their size. Filters small enough to cover a camera lens were measured with the Perkin-Elmer Lambda 7 operating in transmission mode. In order to eliminate contributions from wavelength-dependent polarization changes from the grating monochromator, the in- strument background was normalized with a sheet of polaroid in the sample beam. A second (test) polaroid filter, set to the same transmission vector orientation as the first, was introduced into the cell compartment but closer to the exit window. In this way, spectral intensity changes would be due to light absorption only, as expected for an isotropic light source. Large sheets of polaroid filter were measured using a monochromatic illumination as- sembly composed of a 2.5 kW Xenon arc lamp and collecting optics (Oriel Scientific) and a Monospek 600 monochromator (Rank-Precision). A 1.8-metre-long quartz fibre optic (Schott) conducted the light to the polaroid filter placed in front of a UV/visible