276 JOURNAL OF COSMETIC SCIENCE These minerals, of Moh's hardness 2.5 to 3.5, are softer than tooth enamel (Moh's hardness about 5), thus ensuring that their potential for serious tooth wear in extended use is acceptably low. This is not the case, however, in relation to dentine surfaces commonly exposed when gum recession has taken place. Dentine surfaces can be eroded by these abrasives during brushing, and therefore it is very important to control this property in formulated toothpastes. Products designed for unrestricted home use are necessarily constrained to moderate-to-low dentine abrasion rates defined and measured currently by standard radiotracer techniques (8,9). Fifty years ago, Kitchen and Robinson (5) established the strong positive correlation between the abrasive stain removal properties of dentifrices and the rate of dentine abrasion. The apparent similarity in the mechanical properties of dentine and mature stained pellicle has therefore made it difficult to dissociate dentifrice abrasive perfor- mance from its damage potential. A widely used method of obtaining subjects with stained teeth for clinical studies is to make them brush only with water (5) or an abrasive-free paste (10) for a month or so. Comparative paste performance tests can then be carried out either under controlled brushing conditions provided by a hygienist or within the panelist's own regime. Satisfactory clinical methods for visual assessment of stain and its reduction have been established (11), and an in vitro stain removal screening test using artificially stained bovine teeth has been devised by Stookey et al. (12). This widely recognized method for comparing potential dental abrasive cleaning power is claimed to correlate well with clinical observations of mature stain removal. Accordingly, the performance of a range of pastes tested for us by the Indiana University Oral Health Research Institute using the Stookey method closely followed their respective dentine abrasion rates. For the reasons discussed above, it is not surprising to find in most clinical performance studies that regular toothpaste abrasive systems are not powerful enough to remove mature stain instantly or completely, but are able only to reduce it somewhat (11,13). Moreover, this test protocol hardly represents the reality of everyday brushing, in which up to 24 hours is insufficient time to acquire any visible deposit under normal condi- tions. It is likely, therefore, that the primary abrasive function in dentifrices is not stain removal but stain prevention. The acquired pellicle, composed mainly of salivary gly- coproteins, is known to form readily on tooth enamel surfaces and to thicken and toughen with time (13-15). Undisturbed, it can grow to 10 microns in thickness and become colored (5) over a period of weeks. However, if the young (up to 24-hour) pellicle is substantially removed during each brushing, no build-up of potentially visible films can occur. Inefficient toothbrushing, for whatever reason, will therefore lead to stain. This usually forms in areas not exposed to the regular abrasive action of foods and toothbrush bristles, particularly in the gingival crevice and interproximal regions of the dentition. Under habitual poor brushing conditions, therefore, the abrasive function of the paste is vir- tually redundant in these areas, and neither stain removal nor prevention is likely to be achieved however powerful the paste abrasive system. In the current absence of safe and effective chemical treatments, persistent stainers have the choice of remaining so, vis- iting the dental hygienist for professional cleaning at regular intervals, or improving their brushing technique.

DENTAL STAIN PREVENTION 277 This paper discusses a new evaluation technique for examining abrasive stain prevention by dentifrices, and demonstrates its predictive function in relation to stain prevention studies in vivo. The principal object of the new model was to simulate relatively im- mature 12-to-24-hour-old pellicle films on smooth hydroxyapatite surfaces. MATERIALS AND METHODS LABORATORY STUDIES Substrate. In order to ensure maximum reproducibility consistent with reality within the test, a model hard substrate was employed. Thus highly polished sintered 17-ram hydroxyapatite (HAP) discs (Calcitek Inc, Carlsbad, California) were used to simulate enamel tooth surfaces. This was achieved by lapping the discs using a rotary grinder/ polisher (Buehler UK Limited, Coventry, UK) on wet P600, then P1200 metallographic grinding papers to give a mirror finish. Typical degrees of roughness (Ra) measured profilometrically were of the order 0.1-0.2 microns, similar to that of teeth. Care was taken not to touch these surfaces after polishing. Soil. A self-indicating model organic film intended to simulate the gross mechanical properties of up-to-24-hour pellicle was formed by precipitating an iron (III) complex with tannic acid from aqueous solution directly onto the polished HAP discs. Freshly mixed solutions (0.1% w/w) of diammonium iron (II) sulphate 6-hydrate and tannic acid (GPR grades, BDH, Poole, UK) are initially colorless, but form a dark colloidal iron (III) tannic acid complex ("ferric tannate") on contact with air. The fresh mixture is thus painted onto HAP surfaces with a fine squirrel-hair brush and gently dried with a warm hairdryer. Three thin applications were found to provide a colored organic film of thickness about one micron, which is believed to be of the correct order of thickness to simulate the immature pellicle of interest (14). This film on HAP discs gives a darkness measurement of L* = 50 + 5 (CIE 1976 L*a*b* system) as determined by a Minolta Chromameter © (Minolta Limited, Milton Keynes, UK). Although it is conceded that a ferric tannate film cannot simulate the chemical prop- erties of stain-precursive dental pellicle, it displayed upon inspection many of the key physical properties of such a layer. Thus it proved resistant to brushing with nylon filaments in the presence of all common toothpaste ingredients except abrasive agents. This is consistent with the findings of Kitchin and Robinson (5) and others (11) that regular brushing with abrasiveless systems cannot prevent long-term stain formation. The model soil/substrate additionally proved to be stable, rapidly and easily prepared, and progressively and reproducibly sensitive to different dental abrasive types and slurry concentrations under mechanical brushing conditions. This prompted an investigation of the abrasive performance of a series of paste systems with a view to determining to what extent this might mirror clinical stain prevention data. Mechanical brushing regime. The ferric tannate (FT) stained hydroxyapatite discs were mounted horizontally in the bottom of a trough containing the aqueous toothpaste slurries under test, and weighted toothbrush heads oscillated over the disc surface by way of a mechanical scrubbing machine. A modified Martindale abrasion tester (Goodbrand- Jeffreys Ltd., Stockport, UK) was found to be particularly suitable. In this apparatus the brush heads describe Lissajous' figures, which combine linear and elliptical motions to