Importance of Film
In the open eye, the exposed surfaces of cornea and sclera are covered with a very thin film of tear fluid. This has both protective and nutritional properties; its thickness changes due to evaporation while the eye is held open, and during prolonged eye-opening the film may break up to expose surface epithelial cells directly to the air. The tear break up time (TBUT) is an important clinical parameter in defining the normality of function of the eye. During normal life, break-up will perhaps occur only comparatively rarely, as blinking is very rapid and is a nearly automatic response to symptoms of drying. However, normal life for many people now includes prolonged periods of computer or other visual screen use, and it is known that the blink rate falls when paying close attention. Hence the tear film may break up between blinks. The Corneal Protection Index (CPI), defined as the ratio of TBUT to the length of the interblink, can be used to indicate the boundary between normal and dry eye.

Position and Extent of the Tear Film
The term tear film is normally used to describe the film of fluid covering the corneal surface and contained between the lid margins. In fact we should also include the film overlying the exposed bulbar sclera; however, because this surface is rough and irregular it is much harder to obtain information about its nature by the usual reflectance-based methods used for the cornea, so this area is often ignored. At present we cannot even say with certainty that a uniform and continuous tear film is present over all the exposed sclera. However this area is not negligible in upward gaze it may contribute 60% or more of the total exposed area. It is also neglected because it is of much lower importance than the cornea in the visual process, and because, having its own blood supply, it recovers more readily from injuries and infections.

The exposed area is quite closely dependent on interpalpebral height, which in turn is determined by the direction of gaze (exposure is considerably greater in upward than in downward gaze, as the upper lid follows the movement of the globe). Thus in downward gaze not all the cornea is exposed, along with a small area of sclera, while in upward gaze we may see all the cornea plus a variable amount of sclera both above and below the limbus as well as larger lateral areas. A rough linear relationship between area and palpebral height is often used: Area (cm2) = 0.28 X (height in mm) — 0.44. A more precise value can be obtained by computer analysis of images of the eye, allowing separate estimation of the areas of exposed cornea and sclera. Typical values for the area in normal level gaze are 2-3 cm2, of which 45-55% is cornea. These figures help to strengthen the recommendation to those doing much computer work, to keep the screen as low as possible to minimise ocular exposure and drying because of the reduced blink rate.

The total area of the human conjunctival sac has been estimated as 16 cm2. If all this is covered by a layer of gelatinous mucus with a water content of about 90%, say averaging 1 |xm thick, 1.44 |xl of fluid would be contained. Where the lids overlie the globe, it is possible that two such layers on the apposed surfaces will be in contact, giving an effective fluid thickness of 2 |xm. However, it seems unlikely that in this case there would be free flow of fluid (e.g. fresh tear fluid entering the upper conjunctival sac through the tear ductules) although fluid transport could occur through a 'squeegee' mechanism in blinking.

Formation of the Film
As the lids close during the blink, the upper and lower menisci are pushed ahead of them and sweep up the fluid forming the preocular film, rather like a windscreen wiper. In the opening phase of the blink, the viscosity of the tears causes fluid to be pulled out of both menisci to create a new film, but opposed to this is the negative pressure due to the concave tear meniscus. As long as the lids are moving, fluid is spread, but when the lids become stationary there is within 0.3-1 s a settling down or rearrangement whereby fluid is pulled back into the meniscus while the bulk of the spread film remains intact. The region closest to the meniscus is however considerably thinned and if fluorescein is instilled, a 'black line' can be seen around the rim of the tear film. This line is so thin that it contains too little dye to fluoresce, and it acts as a barrier to diffusion or flow of fluid into or out of the film in the interblink period. Hence the film is effectively isolated from the rest of the lacrimal system while the eye is open, and is subject to different influences such as evaporative loss at these times. The isolated film has been referred to as 'perched' because it covers the exposed eye but is in a sense independent of the ocular adnexa.

Volume of Various Compartments
We can distinguish three distinct components of the fluid in the lacrimal sac: the film itself, lying between the lid margins; the continuous line of meniscus around the lid margins, joining at the outer canthus and around the caruncle, and the fluid under the lids.

It is still not clear what volume of tears lies under the lids, or whether this should be included as part of the tear film. In the normal eye the lid margins glide in contact with the globe during a blink, and it is thought that there is a slight curvature inwards of the margin of the upper lid to give a 'windscreen wiper' action sweeping the film forward as the lids close. This would suggest that the exposed and the under-lid compartments remain separate; but King-Smith et al. discuss the possibility that the two compartments are connected but that during the blink the upper meniscus changes position, being swept down by the advancing lid margin.

Recent experiments on adding saline to severely dry eyes showed that fluid was absorbed (presumably under the lids) before any lid margin meniscus became visible, implying that the two compartments are connected. The mean under-lid volume was calculated to be 5-6 |xl. The volume of tears in the combined upper and lower menisci can be calculated from their total length (about 50 mm) and cross-sectional area, assuming that their profile is a quadrant of a circle; using a mean value of 0.365 mm for the radius of curvature, the normal meniscus volume is about 2.9 |xl. The volume of the preocular film clearly depends on its thickness (see below), but taking commonly-agreed limits of 3 and 10 |xm and an area of 2 cm2, the volume is 0.6-2.0 |xl with a mean probably about 1.0 |xl. Hence the total volume of tear fluid in the external eye is roughly 10 |xl. This does not include additional small amounts such as the fluid over the caruncle.

Clearly there is considerable personal variation in this figure - differences in form of the lid margins, slight inward or outward turning of the lids relative to the globe, positioning of the puncta and height of the palpebral opening can all affect the contained tear volume.

Thickness of Precorneal Film
Estimates of tear volume involve knowing the thickness of the film. This is not easy to measure, although several methods have been used over the years. Simple methods include isolating an area of tear film by pressing the end of a wide-mouthed syringe onto the eye and measuring the volume of fluid sucked off, absorbing fluid over a known area by placing a disc of absorbent paper on the eye, or measuring fluorescence intensity after adding a known amount of fluorescein to the film. More recently the variation of intensity of light reflection has been analysed in three ways (varying angle, frequency or wavelength). Ocular coherence tomography can measure corneal thickness with and without a contact lens and estimate the film thickness by difference. All these methods are summarised by King-Smith et al. Some estimates of tear film thickness by these methods are given in table 1. The film thickness over the anterior surface of a contact lens is generally thinner than the precorneal film, and less stable, although this varies with the contact lens material and depends on factors such as degree of contamination of the lens surface by tear components.

Volume Flow of Tears Into and From the Eye
The volume of tears in the external eye at any moment is a balance between the rate of inflow of fluid from the lacrimal gland, from the accessory lacrimal tissue and by permeation of water from the corneal epithelium through aquaporin-controlled channels. Removal of fluid is principally by drainage through the puncta following each blink, and by evaporation from the open eye. When the lids close, the upper and lower puncta press on each other and prevent outflow, but as the lids open there is a drop in canalicular pressure and fluid is sucked into the puncta from the marginal lacrimal lake [11]. Evidence for absorption of water by the corneal or conjunctival epithelium is lacking, although it is suggested that some or all of that passing down the canaliculi is absorbed before it reaches the nose.

There is considerable variation in the rate of inflow of tears. It has often been suggested that in the quiet eye there is a 'basal rate' of flow, augmented by different degrees of stimulation; one variant is that the basal secretion is produced by the accessory lacrimal tissue (about 10%of the total) and stimulated reflex or psychic tears by the main lacrimal gland, but there appears to be no firm evidence for this.

Another view is that all secretion is stimulated, that in the quiet eye being produced simply in response to opening of the eye. Most clinical estimates of tear flow rate are based on the Schirmer test and its variants; these are described below in 'Clinical Tests'. Published values of the 'unstimulated' flow rate are usually around 1.2|xl/min or roughly 1.2 ml/day (assuming a 16-hour waking cycle, since tear output is largely inhibited during sleep), with a turnover rate of 16%/min. However, using the Fluorotron Master instrument, a much lower value was found of 0.15|xl/min (about 0.15ml/day from each eye) with a turnover rate of 8.2%/min. Stimulated flow rates are much greater - up to 50 or 100 times more; 40-50 |xl in <1 min has been reported with nasal stimulus by ammonia. Since the myoepithelial cells which surround the acini of the lacrimal gland contract in this process, it seems possible that some of the released tears are preformed and the actual secretory process may be somewhat slower than at first appears. It is not clear whether there is a 'maximum' rate of secretion; sustained rates are generally less than the 40-50 |xl/min already mentioned. Regulation of Tear Production
The innervation of the lacrimal gland is complex. The reflex arc is particularly important, involving fibres from the fifth cranial nerve in the cornea, conjunctiva or surrounding tissues. There is also innervation by both the parasympathetic and the sympathetic systems, inducing positive and negative control of secretion respectively. The parasympathetic route indicates some of the complexity: starting from the lacrimatory nucleus in the brainstem of the facial nerve (cranial nerve VII), parasympathetic fibres follow the greater superficial petrosal nerve to the pterygopalatine ganglion; the conventional view is that from there the secretory fibres of the lacrimal nerve follow the zygomatico-cotemporal nerve and join the lacrimal nerve of the ophthalmic division of cranial nerve V and enter the lacrimal gland. However, there is evidence that a number of rami orbitales pass from the pterygopalatine ganglion and some of these travel directly to the lacrimal gland.
The innervation of the accessory lacrimal tissue is even less well known, but it is assumed that it is controlled in the same way as the main lacrimal gland, as they are histologically very similar.

Composition of the Tears in the Conjunctival Sac and Origins of Secretions
Several different collection techniques have been used, but usually collection is from the lower meniscus, or sometimes from the conjunctival surface of the slightly everted lid, or among the folds in the lower fornix. Some workers have used absorbent sponges placed in the lower fornix, which is effective but has the disadvantage of picking up mucus as well as fluid tears. It is still not possible to collect from the actual film, e.g. by blotting the ocular surface, without some damage to epithelial cells and contamination by cellular contents. One should be clear whether the aim is to collect stimulated or unstimulated tears. Stimulation of flow may be by bright lights, a cold stream of air on the cornea, tickling inside the nose or tweaking nasal hair, or by exposure to specific lacrimatory substances such as onion vapour, ammonia or chloracetophenone. If unstimulated tears are needed (for example, for osmolarity measurement), with collection at the slit-lamp, one must avoid passing the light beam across the pupil. We can classify the various components of the secretion as intrinsic or accessory in origin.

Intrinsic Secretions
Intrinsic secretions are produced in the main lacrimal gland (and presumably also from accessory lacrimal tissue since there is no apparent histological difference between the two types of tissue).

Aqueous Component
The aqueous part of the tears forms the bulk of the lacrimal secretion; it is actively secreted, and linked to the secretion of proteins (see below, (Major Proteins)- Although there is some input via aquaporin-controlled water channels in the corneal or conjunctival epithelium, its main source is the lacrimal tissue, where it is produced by the acinar epithelium and collected by the ductules. There is some modification and reabsorption in the ductules before delivery via the main lacrimal ductules to the outer upper fornix. It is possible by everting the temporal portion of the upper temporal lid and by finger pressure prolapsing the lacrimal gland slightly into the fornix to see one or two of the orifices, and if fluorescein is added then clear rivers can be seen in the fluorescing tears indicating the position of their orifices. During sleep or prolonged eye closure, the output of both proteins and water from the lacrimal gland changes (see below, 'Major Proteins').

The rate of secretion of lacrimal fluid varies considerably between the quiet eye and active stimulation (see 'Volume Flow of Tears into and from the Eye'). The ageing lacrimal gland suffers progressive fibrosis and loss of functional acinar tissue so its output gradually falls, creating tear film conditions similar to the earlier stages of the aqueous-tear-deficient form of dry eye.

Salts
Electrolytes are actively secreted by acinar and ductal epithelium of the lacrimal gland, and can be seen from the relative proportions of various ions not to be a serum filtrate. The pH of tears usually lies within the range 7.2-7.6 but may be higher on prolonged eye-opening through loss of C02; the value in neonates is about 6.8. Tears exert a buffering action due to their content of bicarbonate ion, proteins and other components, although the turnover rate has also been shown to be part of the response to pH challenge.

The osmolarity of the tears is determined almost entirely by their electrolyte content, since the molarity of even the major proteins is low in comparison. For normal unstimulated tears the generally accepted value is 302 ± 6mosm • kg1.

Major Proteins
Human tears contain four major proteins (each 15-20% or more of total protein) - lysozyme, lactoferrin, lipocalin and secretory IgA. The protein of unknown function previously referred to as 'tear-specific prealbumin' is now known as tear lipocalin, a member of the lipocalin superfamily of small proteins with lipid-binding properties. There is some evidence for interactions between lipocalin and both lysozyme and lactoferrin. Lysozyme, lactoferrin and lipocalin are secreted by the acinar tissue of the lacrimal gland. The secretory form of IgA, in contrast, is produced by interstitial plasma cells embedded in the gland but external to the acini; the IgA dimer, consisting of two monomeric IgA molecules held together by a J or joining piece, are transported through the acini and the secretory component characteristic of completed slgA is added. Control of secretion of lacrimal gland proteins appears to be linked to that of water: when output of water falls, so also does production of the proteins, and the concentrations of lysozyme, lactoferrin and lipocalin appear fairly constant. During sleep, as mentioned above, fluid secretion declines, and after about 2 h may approach zero. Output of slgA, however, continues as the plasma cells producing this protein are not under the same control as the lacrimal gland, and the same amount of slgA in a greatly reduced volume of aqueous appears as a steep concentration rise. At the same time, polymorphonuclear leukocytes accumulate, with the result that the tear film under the closed lids becomes much reduced in volume, sludgy and turbid, and has been described as being in a state of subclinical inflammation.

IgG and serum albumin are frequently also reported in tears, but since their levels vary with severity of disease or irritation it is considered that these proteins are not normal constituents but indicate leakage from conjunctival blood vessels.
The accessory lacrimal glands, making up 10% of all lacrimal tissue, are distributed at a number of sites within the conjunctiva. They have historically been named as the glands of Wolfring, Krause, etc. but appear to be histologi-cally identical to the main lacrimal gland and to have similar innervation. All the major lacrimal proteins have been identified immunochemically in this tissue. Although the reflex response to irritation is less pronounced, the tissue can produce enough lacrimal fluid to maintain an adequate tear film in the quiet eye even in the absence of the main lacrimal gland.

Accessory Secretions
Several components are added to the aqueous tears within the conjunctival sac, and it is the combination of all these which produces the physiologically functional tear film and influences its formation and stability.

Lipids
A complex mixture of lipids is delivered from the meibomian glands opening on the lid margin at the mucocutaneous junction. These glands are large, tubuloacinar structures lying within the tarsal plate and related to the sebaceous glands of skin; although the surrounding tissue is richly innervated, no specific fast-acting nervous stimulation is known, and they appear to be free-running, secreting lipid continuously. As with sebaceous glands of skin, modification of systemic hormonal status may affect output, but the response is on a scale of months. Compression of the tarsal plate in blinking causes a small amount of oil to be squeezed out of each gland, but repeated heavy or forcible blinking can deplete the supply within the duct of the gland so that delivery is reduced until synthesis catches up with excretion. Conversely, during sleep there is no squeezing of the glands, so the elastic ducts fill up until some critical pressure is reached and excess leaks out onto the closed lid margins, where it either flows or is rubbed away, or forms flakes on the lashes.

In the lid-opening phase after a blink, a fresh air/water interface is rapidly generated, and oil (or at least the more surface-active components) spreads onto the tear film, probably forming a largely monomolecular film. It is thought that this initial spreading is followed by a second phase in which a fluid but less surface-active fraction spreads over the first to produce a multilayered oil film structure. Its thickness can be estimated from its interference colours (e.g. as seen with the Keeler Tearscope®); normal thickness is in the range 40-90 |xm. The surface tension gradient created within the film by this spreading may cause Marangoni flow, pulling aqueous tears from the upper and lower menisci and thickening the overall tear film.

The meibomian oil contains several phospholipids, principally phos-phatidylcholine and phosphatidylethanolamine, which with a small amount of free fatty acids and cholesterol make up the surface-active fraction. The non-polar fraction consists largely of wax esters (fatty acid + long-chain fatty alcohol) and cholesterol esters; branching in many of the acyl chains ensures that the melting range of the mixture is close to lid-margin temperatures. Together, they form a layer shown to retard the evaporation of water from the surface of the tear film. Recently a model has been proposed for the structure of the oil film.

Lipids of non-meibomian origin have also been found in the tears, although reports are still incomplete. A mixture of non-polar lipids, mainly triacylglycerides, a small amount of phosphohpid, and a substantial proportion of unidentified glycolipids has been described. Since no free lipids are found in tear fluid, it is presumed that these are bound to lipocalin, which is the only major protein with strong lipid-binding characteristics.

Mucins
Mucins are a complex class of glycoproteins with a very high carbohydrate content; their main characteristic is the bottle-brush structure of a polypeptide backbone with many tandem repeats of amino acid sequences and a high proportion of serine, threonine andproline, with a large number of ohgosaccharide side chains O-glycosidically linked to Ser or Thr. They are the products of the family of MUC genes, and are of two main types: secreted or 'soluble' mucins of which the most important in the eye is MUC5AC, a gel-forming mucin produced in conjunctival goblet cells, and epithelial mucins, where the polypeptide backbone has a membrane-spanning region anchoring it to the plasma membrane of epithelial cells of cornea or conjunctiva, such as MUC1. The epithelial mucins (principally MUC1, 4 and 16) form the glycocalyx visible in transmission electron micrographs of the ocular surfaces, and a major function appears to be the anchoring of a gelatinous layer of secreted mucin so that a lubricating layer is present on all the surfaces gliding over each other during blinks or ocular movements. Mucins typically contain more than 50% carbohydrate, and water makes up more than 90% of mucin gel.

Minor Components and Small Molecules
Tears contain a large number of small molecules and minor components which can protect the corneal surface or which are produced in response to specific conditions such as inflammation. Defensins are a family of small proteins (Mr about 8,000) with antimicrobial properties (see below, 'Antimicrobial Protection'). Several cytokines associated with inflammation (IL-la and IL-1 (3, IL-6 and IL-8) have been identified in normal tears. However it is not always clear whether these factors are derived from the lacrimal gland or secreted by the conjunctival epithelium, or by leakage from the surrounding blood vessels. Enzyme activity of various kinds can be detected in tears, although the amount of the appropriate protein may be very low. Thus, catalase, superoxide dismutase, and glutathione peroxidase have been reported, among others, and are presumed to have an antioxidant protective role.

A number of systemic drugs can be detected in the tears. The actual source (conjunctival vessel leakage, transport through corneal epithelium, or lacrimal gland secretion) is not always obvious. If corneal, this could imply a specific membrane-associated transport mechanism, or an ability to pass the tight intercellular junctions, and this latter is thought to be related to the lipid solubility of the drug. Thus, phenobarbital, carbamazepine and methotrexate, which all have reasonable lipid solubility, have been detected in tears at levels comparable to those in serum, whereas ampicillin is less lipid-soluble and is found only at a very low level compared to serum. Acetaminophen is excreted in the tears at comparable levels to serum. It is known that systemic cytosine arabi-noside can cause keratitits, and this is thought to follow secretion into the tears. Rifampicin and its metabolites appear in tears, which may be coloured red-orange and cause staining of contact lenses.

Functions of the Film

Nutritional Aspects
Because of the requirement for transparency, the cornea has no blood supply. Delivery of gases and nutrients by diffusion from blood vessels at the limbus would be too slow, so these are supplied directly from the tear film; the film acts as a coupling medium for oxygen from the air (as is clear from the comparative performance of contact lenses with differing Dk values). A similar function takes place on the endothelial side of the cornea from the aqueous humour of the anterior chamber. In the open eye the tears, being in contact with air, are assumed to be saturated with oxygen (i.e. 155mmHg); however, when the eye is closed, oxygen must be supplied by diffusion from the blood in the conjunctival vasculature (55mmHg), so the metabolic status of the corneal epithelium changes markedly between the two states. It should be noted that in the closed eye the coupling medium to the cornea must actually be the film of tear fluid filling the space under the lids. The thickness of this is not exactly known, and it is also assumed that there is a relatively thick mucous layer filling most of this space, caused by the apposition of the mucous layers covering both cornea and tarsal conjunctiva.

The tears transport oxygen to the corneal epithelium, and remove metabolic carbon dioxide. Comparatively few other nutrients are found in significant quantities. It is suggested that glucose is supplied to the cornea entirely from the posterior or endothelial side, and that the corneal and conjunctival epithelia are impermeable. Tear glucose levels are low, and little changed in diabetics; reports of higher levels may be due to local tissue damage and assay of released glucose. Lactate and pyruvate are also found, indicative of the metabolic activity of corneal tissue. The growth factors EGF and TGF-a have also been detected.

Protective Roles

These can briefly be classified in two distinct areas:

Physical Protection
Many threatening or noxious attacks on the eye are averted by the rapid blink reaction, or by aversion (head turning or brow lowering); some lighter invading materials such as airborne dust, hairs or bacteria may be reflected from the surface of the tear film, especially hydrophilic particles which have been observed to bounce off the oil film. The mucous gel coating of ocular surfaces traps, absorbs and immobilises many particles and microbes, and removes them from the eye as part of the mucous thread which is swept down into the lower fornix and eventually extruded onto the skin of the inner canthus. The lubricating action of the mucous layer also prevents shearing damage to the surface epithelium at the high speeds (as high as 20 cm/s) achieved during the blink.

Lipocalin is the principal lipid-binding protein in tears, and a role for this protein has been suggested in scavenging excess lipid from the ocular surface or the surface of the mucous layer to avoid the development of non-wettable patches that would lead to tear film break-up. As yet this has not been supported by analytical studies.

Antimicrobial Protection
Several of the components of tears have antimicrobial functions. Lysozyme is well known for its muramidase activity in the outer cell wall of Gram-positive bacteria, while both lactoferrin and lipocalin have iron-sequestering properties which inhibit siderophilic bacteria. Secretory IgA exerts immunological protection after priming of the plasma cells against specific microorganisms and viruses; priming can be via mucosa-associated lymphoid tissue in the conjunctiva or elsewhere. Recently a group of small protective peptides known as defensins have been identified in the tears by immunochemical means. Several members of the a and (3 families of defensins were identified in normal tears, lacrimal gland, and inflamed conjunctiva. These have a broad spectrum of antimicrobial activity (bacteria, fungi and viruses) and are claimed to accelerate epithelial healing.

All these factors need to be considered in relation to ocular surgery, especially physical aspects such as the placement of inflow ductules and puncta/ canaliculi for drainage, avoidance of distortion of surface or conformation of lids on globe, and the removal or remodelling of conjunctiva.

Structure and Stability of the Precorneal Film
Many structural models of the preocular tear film have been proposed over the last 50 years. These are mainly based on the three-layered structure of Wolff, which has a layer of gelatinous mucus in contact with the epithelial surface (since modified largely on the basis of electron-microscopical evidence to include the surface glycocalyx), the bulk of the thickness made up of an aqueous solution of the proteins and other water-soluble molecules, and a surface layer of meibomian oil. More recently, a model has been proposed for the rat involving only two layers, in which the bulk of the film was aqueous/mucous plus an oil layer, with no differentiation into separately identifiable aqueous and mucous layers. A somewhat similar model is suggested for the mouse. It is not clear whether either of these models should also be expected for the human, and space does not allow an extensive review of the aspects of all the available models. Despite much work on the human tear film and in many species of animal, we have not yet arrived at one consistent model which can satisfactorily explain all aspects of formation, stability and function of the film.

In view of the nutritive and protective properties of the tear film, it is clearly desirable for it to cover the exposed surface of the eye throughout the eye-open period between blinks. Evaporation can be measured, but we should remember that most evaporative loss will be from the film, while the bulk of the available fluid is in the menisci or under the lids, and it is from these compartments that samples are collected for analysis.

Hence local changes in osmolarity may be greater than usually thought, and corresponding effects on film stability may be masked.

The main test of tear film stability is the break-up time (BUT), i.e. the time taken after the last complete blink for signs of rupture and dewetting of the film to be detected. Tests differ in whether they are invasive (instillation of fluores-cein to show break-up as black spots, FBUT) or non-invasive (detection of distortions of the reflected image of a grid from the cornea, NIBUT), and the value taken to indicate the borderline between normal and unstable or dry eye may vary according to method: between 5 and 180 s (ca. 5-20 s for FBUT, or ca. 10-30 s for NIBUT). However, other factors such as number of repeat measurements, time of day or racial characteristics of the subject can also influence the outcome. Perhaps the most reliable use of BUT is in assessing the effectiveness of clinical treatment. FBUT is widely considered to have poor repeatability, although this may depend on the quantity of fluorescein introduced, since this can itself affect tear film stability. NIBUT also shows considerable variation when the same subject is measured on successive days, and also for repeated measurements on the same day, although this may be due to increased tearing in response to holding the eye open for long periods. Nevertheless, BUT is a valuable guide to tear film stability. It is considered satisfactory to take the mean of three successive measurements (in the case of FBUT, adding as little fluorescein as possible).

Tests on Tears
There are many tests which can be used to assess tear film composition or function. These may be classified as subjective, where some element of judgement is required on the part of the observer, such as in grading the extent or severity of a sign, on some predetermined scale such as 0-4, as -/+ or + to + + + ; or objective, involving use of methods or equipment capable of giving a more precise value. A further division is between those tests which can be carried out under clinical conditions (although the results may be interpreted elsewhere), and those where samples are examined in the laboratory or the patients themselves are examined outside the clinic.

Clinical Tests
Apart from the basic tear break-up test, these include estimating tear volume from the Schirmer paper strip test. The recommended Schirmer strip is of Whatman No. 41 filter paper 5 mm wide and 35 mm long, with the terminal 5 mm bent to hook over the lid margin. The test can be applied in various forms, which measure different aspects, but the nomenclature is confusing. The test can be with or without anaesthetic. Schirmer's original test (Schirmer I) is without anaesthetic and does not include stimulation, other than that due to the inserted paper. The 5-min wetting length is taken to represent the basal unstimulated flow. A wetted length of 15 mm or more is taken to indicate normal production. A variant of this is the Jones test which also measures the basal rate, but uses anaesthetic and is carried out in subdued lighting conditions to minimise reflex tearing. The normal response is a wetted length of 10mm or greater. If the basal rate is normal but the reflex response to stimulation is thought to be defective, Schirmer II can be applied, which uses anaesthetic but includes stimulation of reflex tearing by nasal irritation with ammonia vapour, onion vapour or a cotton applicator. A reading of 5 mm or less in 5 min is indicative of aqueous-deficient dry eye. There are numerous variants of the original Schirmer tests; despite many reservations about its meaning, it is still generally accepted that it gives useful information. The cotton-thread test is a variation of the Schirmer test, using a loosely-twisted thread which is less irritating to the eye (and less likely to provoke reflex secretion); the steady-state output of the lacrimal gland is being assessed, whereas without anaesthetic (and hence with the irritation of insertion of the paper) the reflex response of the lacrimal gland is probed. The disadvantage of the thread method is that because it does not provoke tearing, it measures only the fluid already available in the conjunctival sac.

The normality of tear volume is also estimated from meniscus height or meniscometry where meniscus curvature is calculated from reflection of a striped target.

The measurement of evaporation itself is possible under clinical conditions, but no commercial instrument exists. One instrument, which calculates evaporation rate from the rate of rise of humidity inside an eyecup, is currently used in assessment of dry eye patients in the clinic.

The use of the Fluorotron Master to measure turnover time or clearance rate of tears from the eye has been mentioned above in Volume Flow of Tears Into and From the Eye.

The thickness of the lipid layer is assessed from the interference colours seen on reflection of light using an instrument such as the Keeler Tearscope. Meibometry, in which the lid margin is blotted with a tape and the change in transmission of the tape due to the oil picked up is measured, can give information about the availability of oil, and if the lid margin is first cleaned of oil, about delivery from the glands. This is in fact the only currently available objective measure of meibomian gland output.

Laboratory Tests
These are generally more time-consuming or involve the use of more complex equipment than clinical tests. Samples of tears or other secretions must be taken, paying attention to the collection site or conditions. Thus the protein composition may be analysed by high-performance liquid chromatography, although this will show only the major proteins and not minor components, which may have to be detected by assaying collected column effluent fractions for enzyme activity or other functions. Alternatively, polyacrylamide gel electrophoresis can give detailed information about the protein composition of tears.

Tear osmolarity is a good indicator of high rates of evaporative loss, and can be measured on collected tear samples. The Clifton nanolitre osmometer (depression of freezing-point principle) is still considered the gold standard method despite its many practical difficulties; the Wescor vapour pressure osmometer is simpler and could be used in the clinic, but may have a considerable reading error with tear samples < 1 |xl, which one must use to avoid reflex lacrimation and dilution of the tear film during collection [59]. A simple, rapid and very sensitive commercial instrument is promised, but was not available at the time of writing. Tests of Quality
Whereas one can, by detection of deviations from normal composition, conclude that the tears are of less than the required quality to maintain stability and function, it is much harder to devise tests to establish whether the performance of a sample of whole tears is of the required quality. Perhaps the only such test is tear ferning. A small sample of fluid tears (about 2 |xl) is placed as a droplet on a clean microscope slide and allowed to dry, then examined under the microscope . Viewed at X50 to XI00, feathery patterns of salt crystals are seen, and the degree of complexity of these correlates well with other measures of tear quality or performance. Although often called the 'mucus ferning' test, it is in fact less dependent on mucus content than on the balance of electrolytes, but much more exploratory work needs to be done before it can be considered altogether reliable.
Compositional tests as indicated in 'Laboratory Tests' can be applied to show that some assumed best or 'normal' assembly of components is present. But this is complicated in that it changes to some extent with age or other physiological states (e.g. the menstrual cycle). Vital staining can also give information about the completeness of the film. Thus, in the same way that fluorescein is used to indicate breaks in the epithelial surface, staining with rose bengal is considered to depend on breaks in the mucous layer covering the ocular surface, revealing the unprotected and presumably unlubricated epithelial surface beneath.

Physical properties such as viscosity or surface tension can be measured if adequate volumes of tears are available, and can indicate the normality of the tears.

Conclusions
The normal tear film is metabolically functional, protective, and nutritive. Problems arise if its stability is compromised by anatomical factors such as the improper meeting of lids or the closeness of their fit to the globe, blockage of the drainage routes, surface roughness or epithelial damage. Inflammation involves the secretion into the conjunctival sac of many additional components, of both tissue and serum origin, and these can materially alter the physiological functioning of the tears. These factors must all be taken into consideration in planning surgical procedures.