26 R. Heywood and R. W. James sodium hydroxide washing enhanced the initial eye irritation. With strong acids and alkalis, washing has no effect. The response of the conjunctiva, cornea and iris are graded on a 0-4 scale. It is at this point that observer bias comes in. The reactions are best scored by technical staff trained to a common standard. Bayard and Hehir (5), investigating reader-scoring variability, found that readers were consistently able to produce a positive or negative reading more than 90•o of the time. If clinicians, be they veterinary or medical, examine reactions, they are biased by their experience of clinical conditions in the eye and tend to score to lower grades than technical staff. Ballantyne et al. (6) suggested grading for lachrimation, blepharitis, chemosis, hyperaemia, sloughing, iritis, keratitis and for corneal vessels. It is common practice to manipulate the scores in order to express the total irritancy as a single number. Attractive as the idea may be to express a complicated biological situation as a single number, the practice must be viewed with caution. Ballantyne and Swanston (7) state that it is not possible to grade eye irritancy simply on the grounds of numerical values and that observed effects must be summarised verbally. There is considerable variability between individual animals, whether one is working with rabbits or monkeys. To reduce the number of animals would decrease the ability of the test to differentiate degrees of irritation, but increasing the group sizes would not increase precision. Comparison of the total Draize score and scores from the cornea only suggest that the conjunctiva makes the initial response to irritation and that the reaction persists for longer than it does in the cornea. The effect of this conjunctival response is not under- stood, but it could be considerable. The present trend in Draize testing is to reduce concentrations of the working solutions in the hope of being able to measure conjunctival response only. Despite our concern about the Draize test, Marzulli and Ruggles (8) reported that reliable and reproducible results could be obtained in different laboratories in distin- guishing an irritant from a nonirritant. They recommend a simple pass/fail criterion, monitoring four parameters. SLIT LAMP EXAMINATION In our experience, this examination, together with fluorescein staining, is the most sensitive indicator of corneal damage in the monkey. If fluorescein is instilled into the conjunctival sac of the rabbit, in nearly all cases staining patterns will result. Kikkawa (9) described intervals of 1-11 days between light and intense staining and correlated the pattern in both eyes. The stain pattern was attributed to physiological desquamation of the corneal epithelium. In our experience, this pattern of intense and light staining has not been established in the rabbit. The turnover rate of epithelial cells in the cornea is high and the vulnerability of the corneal epithelium to insult may reflect the different stages in the life-cycle of the cells. Nevertheless in monkeys this high turnover of cells is achieved without a break in the continuity of the epithelium. The fact that the corneal epithelium of the rabbit is not intact will generally increase the rate and amount of penetration, this being particularly true for water soluble and polar compounds. CORNEAL THICKNESS Corneal thickness can be readily measured following the technique defined by Mishima and Hedbys (10) using the Haag-Streit depth measuring attachments. Burton (11) must

Objectivity in assessment of eye irritation 27 be credited with first realising the usefulness of this parameter in assessing irritancy. Initially, it appeared to be a sophisticated and objective measure of irritancy, free from observer bias, but unfortunately, with experience the limitations become apparent. It is important that the measurement of thickness is made at the apex of the cornea. It is unfortunate that the cornea is not uniformly thick, nor does it uniformly respond to irritation. If sloughing of the epithelium occurs the measurement becomes meaningless. There is again an observer bias as to the point on the cornea which one chooses to measure. Nevertheless, it has been shown with commercial shampoos, Burton (11), and a variety of industrial solvents, Conquet et al. (12), that a highly significant correlation can be shown between the total Draize score and the corneal thickness measurement. It must be concluded that this measurement of corneal thickness can only supplement the clinical observations. INTRA-OCULAR PRESSURE Ballantyne, Gazzard and Swanston (13) showed that a rise in intra-ocular tension is one of the earliest indications that a compound is producing eye irritation. Furthermore, they suggested that there might be a relationship between the rise in tensiola and the severity of the subsequent reaction. There are two indirect methods of measuring intra-ocular pressure. Firstly, there is the indentation tonometer, by which a weighted plunger is applied to the cornea, the most common piece of apparatus used being the Schiotz tonometer. The second and most applicable technique is that of applanation tonometry, which measures the force required to produce a degree of flattening. The use of the Schiotz tonometer is often impractical, because swelling of the cornea cushions the tonometer footplate and a satisfactory measurement cannot be made. To date, there is little published data to support the use of applanation tonometry for measuring ocular irritancy but current investigations - Heywood and Walton (14) - will indicate the potential of this technique. CORNEAL CURVATURE Measurement of corneal curvature is precluded by the effects of irritants on the pre- corneal film, because any disturbance of this film and the superficial epithelial cells of the cornea prevents lining up of the mires of the ophthalmometer and making accurate measurements. CONJUNCTIVA• AND CORNEAL WEIGHT Conjunctival and corneal weights were found to be precise measurements of irritation by Laillier, Plazonnet and Douarec (15), who investigated many organic solvents. For control animals, the percentage dry weight was 26-4•o-½0'8•o, but even with the most severe irritants, this never fell below 10•o. The water content of the cornea does not show a direct relationship to opacity, Wright, Ulsamer and Osterberg (16). EFFECTS ON CAPILLARY PERMEABILITY Changes in capillary permeability in the conjunctiva and the blood/aqueous level can be demonstrated using the Evans blue dye technique. Laillier et al. (15) have shown change in capillary permeability to be one of the first events in eye irritation.