1.2.1. Introduction

There are five main sectors in the glass manufacturing industry: flat glass, containers and pressed ware, art glass, special glass (e.g. optical, ophthalmic, electronic) and fibre glass. Fibre glass was considered in a previous monograph (IARC, 1988) and is not included here (see General Remarks, p. 36, for evaluations). Figure 1 illustrates the basic principles of glass manufacture. The modern production of flat glass is the most highly automated and usually involves tank melting with continuous feeding of batch ingredients and the float (Pilkington) process (Grundy, 1990) for forming. The production of containers and similar products has also become increasingly mechanized. Modern production techniques involve tank melting with continuous feeding of batch ingredients and mechanical blowing or pressing of molten glass. The production of art and special glass can involve a variety of modern and traditional techniques, including manual loading of melting pots and mouth blowing. The production processes currently used to produce glass, other than fibre glass, are discussed below.

Fig. 1

Processes and materials involved in the manufacture of glass

1.2.2. Raw materials

Glass has some of the physical properties of both liquids and solids: when cooled from the hot molten state, it gradually increases in viscosity without crystallization over a wide temperature range until it assumes its characteristic hard, brittle form; cooling is controlled to prevent high strain (Boyd & Thompson, 1980).

Any mixture with those physical properties is theoretically a glass. Glass has a very large number of chemical compositions, which fall into four main types:

1. Soda-lime-silica glasses: These are the most important glasses in terms of quantity produced and variety of use, including almost all flat glass, containers, low-cost mass-produced domestic glassware and electric light bulbs.
2. Lead-potash-silica glasses: These contain a varying but often high proportion of lead oxide (see IARC, 1987a). For example, in the production of heavy crystal glass, the glass batch contains about 30% lead (Andersson et al., 1990), while in semi-crystal glass production the amount of lead is less than 10%. Fabrication of optical glass and hand-blown domestic and decorative glassware depends on the high refractive index and ease of cutting and polishing leaded glass. The high electrical resistivity and radiation protection of leaded glass are important in electrical and electronic applications.
3. Borosilicate glasses: Borosilicate glasses, with low thermal expansion, are resistant to thermal shock, making them useful for domestic oven and laboratory glassware (Cameron & Hill, 1983).
4. Other glasses: There are many other glass-forming systems, with a variety of compositions.

The raw materials for glass are mainly: silica (see IARC, 1987b), in the form of sand or crushed rock quartz; soda ash, or in some cases salt cake (sodium sulfate); potassium carbonate or nitrate; crushed limestone or dolomite; red lead or litharge; boric acid or borax; and cullet, which consists of broken or crushed glass (Cameron & Hill, 1983).

The major proportion of almost all glass batches is silica sand. Waste glass, or cullet, is an almost universal addition to the batch melted in glass furnaces. The advantages of this raw material are that it facilitates the melting of other components, requires up to 25% less heat to melt than an isochemical batch of raw materials and normally reduces dust during the batching. Batches with optimal melting efficiency may contain about 30–40 wt% cullet. Recycled bottles constitute about 15% of the cullet used in the container glass industry. This amount is expected to increase with the increasing introduction of recycling laws. Raw materials are selected according to purity, consistency, grain size, water content, supply pollution potential, ease of mixing and melting and cost. The commonest ingredients in glass production are listed in Table 1; not all of the constituents are contained in every type of glass.

Table 1

Raw materials used in glass production.

Contaminants and other chemicals are added in small amounts for special purposes. The primary contaminants of sand used in glass-making are ferric oxide (see IARC 1987c) followed by titanium dioxide (see IARC, 1989a), zirconium dioxide and chromium oxides (see IARC, 1990a). The acceptable limits for ferric oxide in glass are 0.005–0.03 wt% depending on the product. Oxides of transition metals, such as copper, cobalt (see IARC, 1991), nickel (see IARC, 1990b), vanadium and tungsten, are also often present and normally not desired.

Chemical or physical decoloration of such glasses is therefore important Chemical decoloration is used to oxidize iron to its weakly coloured trivalent state by increasing the oxygen content of the melt, usually by adding oxygen-producing compounds such as potassium nitrate and sodium sulfate. Physical decoloration is often used when the melt contains a higher concentration of iron. In this case, the addition of other colouring agents compensates for the yellow-green colour of the glass. Common compensating agents are selenium (see IARC, 1975) and oxides of manganese, cobalt and nickel.

Conversely, when coloration of glass is desired, oxides of transition elements are the main colouring agents used. Common transition-metal oxides used in the glass industry include cobalt, nickel, chromium, iron, manganese, copper, vanadium and cadmium Concentrations are typically 0.5–5 wt%. Oxidation conditions affect the colour achieved.

Other chemicals may be added for different purposes: e.g. fluorides to reduce viscosity and aid melting, zirconium dioxide to raise the softening-point and cerium oxide to stabilize glass against ultraviolet discoloration.

1.2.3. Mixing and melting

The three steps involved in the production of glass are melting, fining and homogenization. The cold batch is placed in the tank and melted at 1200–1650 °C. Decarbonation desulfurization and dehydration are the first chemical processes used, followed by partial melting. Silicate formation occurs at about 500–800 °C.

Fining is the process of removing bubbles from the glass melt. This is accomplished in two ways:

• by the addition of chemical fining agents. Sodium sulfate or arsenic (see IARC 1987d) and antimony trioxides (see IARC, 1989b) are generally used. Typically, about 0.3 wt% arsenic oxides or arsenates are added to the batch, although for some glass types up to 1.5 wt% arsenic trioxide may be used; and
• by varying the melt temperature, either by raising the tank temperature with an attendant decrease of melt viscosity and increase in bubble expulsion, or by lowering the temperature and increasing gas resorption.

Homogenization of the batch is generally assured by mechanical stirring.

Accuracy and precision in weighing the various batch constituents are important, as are thorough mixing of the ingredients and prevention of subsequent segregation. Batch handling systems vary widely throughout the industry, from manual to fully automatic. Wet mixing and batch agglomeration-pelletizing, briquetting and compaction are becoming popular, especially for the addition of lead and arsenic compounds.

1.2.4. Pot processing

The older pot process now serves mainly for the manufacture of high-quality glass, such as optical glass, and for small quantities of special glass, such as hand-blown crystal. The pots vary in size, up to those capable of holding nearly 2 tonnes of ingredients. The pots are made from a number of different types of clay combined with flint or silica flour. Pot melting of glass often involves the hazards inherent in hand shovelling and filling. The pots are tempered slowly at 900 °C by electrical heating, fired at 1200 °C and vitrified at 1400 °C in the pot furnace. Open pots containing 150–1000 kg of melt have been used, especially in Europe, to manufacture coloured and optical glasses as well as crystal glass. Closed pots surrounded by refractory brick walls are the system of choice in the USA. The largest pot furnaces could contain 8–10 pots, each with a diameter of 1 m. Special, valuable products (e.g. specially coloured glass) may be produced in tanks built from refractory brick and heated directly, producing 1–5 tonnes of glass daily. This process is particularly well suited for the manufacture of soft (i.e. low viscosity) glass, such as crystal or soda lime glass, which require special production techniques. Modern techniques for special glasses may involve the use of induction heating with the melt contained in platinum crucibles; this technique also involves manual handling of batch materials.

1.2.5. Tank processing

Large volumes of glass are produced exclusively in tanks capable of holding up to 2000 tonnes of melt. Producers of flat glass have always used the largest tanks, formerly for the vertical drawn sheet process and now for the rolled plate and float processes. The tank is divided into two sections, either by a permanent bridge wall or by floating refractory baffles. The larger section into which the fill (i.e. the batch with cullet) is introduced is called the melting end. Most of the tanks provide for enclosed continuous feeding of batch ingredients. The other section, in which the melt starts to cool towards its working viscosity, is called the working or refining end. Partitioning of the tank allows continuous production of glass. The fuel usually used is oil or natural gas.

1.2.6. Forming/making

Glass objects are formed from molten glass by blowing, pressing, drawing and casting. ‘Hand-blowing’ is the classic method of glass fabrication and is used in modern glass plants only for the manufacture of art objects, large objects in low demand that cannot be fabricated by machine, and complicated technical shapes. Partial automation is being introduced into hand-forming operations, where the glassblower blows gobs of the proper viscosity into moulds.

Deep items (e.g. bottles) are formed mechanically by blowing glass into moulds with compressed air. Hinged moulds, which can be opened to remove the ware, are generally used for the blowing operation. These moulds are lined with a water-absorbent coating, which develops a steam cushion between the coating and glass during blowing. A second type of mould is the hinged hot-iron mould, in which the glass comes directly into contact with the mould surface on blowing. Narrow-mouth containers are manufactured in a two-stage process.

Although blowing is used in the manufacture of deep items, pressing is used for relatively flat items, such as dinnerware, optical and sealed-beam lenses, filter glass and television tube panels. Press moulds are commonly made of cast iron, bronze steel or superalloys with galvanized surfaces. Moulds (including hand moulds) are sometimes lubricated with mineral oils (see IARC, 1987f), graphited oils or, occasionally, other organic materials, such as tallow and oleic acid. Hand pressing is now being replaced by machine pressing.

The traditional process for producing sheet glass is to draw it from the furnace by a vertical process which gives it a fire-finished surface. Owing to the combined effects of drawing and gravity, some minor distortion is inevitable. The plate glass passes through water-cooled rollers into an annealing oven. Surface damage must be removed by grinding and polishing. Patterned glasses are still made by this method.

The vertical drawn process has largely been replaced by the float process for flat glass. The Pilkington float process has made possible the manufacture of a glass that combines the advantages of both sheet and plate. Float glass has a fire-finished surface and is free from distortion. Molten glass floats from the huge melting tank (up to 2000 tonnes of glass) along the surface of a bath of molten tin, in which an inert atmosphere of nitrogen prevents oxidation. The glass conforms to the perfect surface of the molten tin. As the glass passes over the tin, the temperature is reduced until the glass is sufficiently hard to be fed on to the rollers of the annealing oven without marking its under-surface. Sulfur dioxide may be applied at this point to reduce further the risk of marking. After annealing, the glass requires no further treatment and is ready for automatic cutting and packing. The advantage of the float process, apart from the mirror quality of the glass surface, is its high production capacity.

1.2.7. Annealing and other processing

Annealing of a glass product is nearly always necessary after any forming operation. The glass is heated uniformly to a temperature sufficiently high to relieve any internal stress, without causing deformation of the object under its own weight. The ware is subsequently cooled slowly to prevent formation of new stresses. Glassware is normally annealed in long, continuous ovens called lehrs, which are usually heated electrically.

After production in different processes, glass may undergo secondary processes, including cutting, grinding, polishing, heat processing, chemical treatment and surface coating. Workers may be exposed to chemical agents during grinding, polishing, chemical treatment and surface coating.

The purpose of grinding is to remove the upper layer of the glass surface. An abrasive produces a small check or crack in the glass, the dimensions of which depend on the grain size of the abrasive: increasingly finer abrasives produce a relatively smooth surface. Natural abrasive grits (e.g. quartz, sandstone, corundum, garnet and diamond) are used as well as synthetic grits (e.g. silicon carbide, aluminium oxide and boric oxide). Water or a suitable cutting fluid is used in grinding.

Polishing is done either mechanically or chemically. Mechanical polishing requires finely powdered abrasives such as rouge (ferric oxide) and cerium oxide. These abrasives operate on the same principle as those in grinding. Chemical polishing includes acid polishing and flame polishing. In acid polishing, the glass may be submerged in a mixture of hydrofluoric and concentrated sulfuric acids or in other acids; hydrofluoric acid is also used for etching glass.

Vitreous enamels are applied by spraying, silk screening and the application of decals. They may be fused to the ware at high temperature. Metallic coatings are produced either by applying a liquid suspension of metal to the surface and then firing it or by evaporation of metal on the glass. These coatings are used on flat glass as a means of controlling transmission and reflection of light, as electrical semiconductors in tin oxide films, as resistors in electronic circuits, as defrosting agents and in aircraft glazing. Lacquers are applied to glass for decorative purposes.

1.3.1. Introduction

A number of potential occupational health hazards are present in the glass industry including silica dust and certain metallic compounds. Exposure to silica has been evaluated previously (see IARC, 1987b).

The production of glass involves the use of many metals, usually as oxides; nearly every element in the periodic table has been used in modern glass technology (see section 1.2). In particular, exposure to lead has been noted in the past: In the production of heavy crystal glass (about 30% lead) and semi-crystal glass (< 10% lead), lead is an important source of occupational exposure. Other elements to which workers are exposed include antimony arsenic, cadmium (see monograph, p. 119), manganese, selenium, nickel and chromium (Andersson et al., 1990); they may be exposed to either dusts or fumes. Glass-blowing offers special opportunities for exposure to the metallic compounds in glass and alloy materials which may migrate up the pipe into the worker's mouth. In Sweden, samples from blowpipes contained lead, manganese, nickel (Andersson et al., 1990; Wingren & Englander 1990) and arsenic (Andersson et al., 1990).

For chemical polishing and matting of glass surfaces, sulfuric (see IARC, 1992) and hydrofluoric acids are used. Some processes also involve potential exposure to polycyclic aromatic hydrocarbons (see IARC, 1983) and asbestos (see IARC, 1987e). The working environment around a glass furnace is hot, and most of the heat is in the form of radiant energy. Significant heat problems also arise during maintenance and emergency repair work: temperatures in areas where men do routine maintenance is frequently in the range of 120–160 °C; under emergency repair conditions it may reach 200 °C (Cameron & Hill, 1983). In the past, asbestos was used as a thermoinsulator in hot structures, as well as in protective clothing. The workers who were exposed to the largest quantities of asbestos were probably construction and maintenance workers rather than those engaged in the actual production of glass (Lucas, 1981; Sankila et al., 1990; Kronenberg et al., 1991).

Very few surveys involving ambient and biological monitoring of workers exposed to chemicals have been reported from the glass industry, and almost all the quantitative data available to the Working Group were for exposure to lead and other metals. In particular, no quantitative data on past exposures were available for the mortality and cancer morbidity cohort studies carried out in Italy and Finland (Cordioli et al., 1987; Sankila et al., 1990).

1.3.2. Exposure to lead

Air measurements in glass manufacturing industries showed concentrations of < 0.001–0.11 mg/m³ lead, with the highest levels in a heavy crystal glassworks (Andersson et al., 1990; Wingren & Englander, 1990).

Because blood lead levels give better information about total lead exposure than lead concentrations in ambient air, biological monitoring was chosen for the evaluation of individual exposures in the studies described below.

Between 1970 and 1973, 49 workers in three Finnish crystal glass plants were investigated for blood lead: The median concentration was [410 µg/L] (range, 120–820 µg/L). At that time, the recommended limit for blood lead in Finland was 700 µg/L (Tola et al., 1976).

Exposure to lead and other substances was evaluated among workers grinding, polishing and glueing leaded crystal glass art objects. Five of six measurements in the breathing zone in the grinding department exceeded the US National Institute for Occupational Safety and Health (NIOSH) criterion of 50 µg/m³. The average of all six samples was 60 µg/m³ (range, 30–80 µg/m³); however, no blood sample contained more than 400 µg/L. The mean blood lead concentration in the grinding department was 291 µg/L (range, 220–360 µg/L) (Gunter & Thoburn, 1985).

Lüdersdorf et al. (1987) examined blood lead concentrations in a group of 109 male workers at two glass-producing factories in Germany. The group was divided into four subgroups, according to their specific activities: melter, batch mixer, craftsman and glass washer. The medians and ranges for blood lead concentrations are given in Table 2, which shows that the highest values occurred in blood specimens from batch mixers.

Table 2

Concentrations of lead in blood in workers in two German glass factories: comparison of subgroups.

Schaller et al. (1988) studied blood lead levels, determined by flame atomic absorption spectrometry, in the German crystal glass industry for the period 1978–86. A total of 6080 determinations were performed for 1625 men and 269 women. The median concentrations (and 68% ranges) varied from 470 µg/L (275–665 µg/L) for men to 255 µg/L (115–415 µg/L) for women ≤ 45 years of age and 270 µg/L (135–455 µg/L) for women > 45 years of age. Table 3 presents the blood lead concentrations in relation to period of examination; it shows a general decrease in internal exposure with time. For men, 764 (14%) of the determinations exceeded the German ‘biological tolerance value’ of 700 µg/L blood. [The Working Group noted that the large decrease between 1978 and 1979 may have been related, in part, to the phasing out of lead in gasoline in Germany and the resultant decrease in environmental exposures.]

Table 3

Blood lead concentrations (µg/L) in the German glass industry for the period 1978–86.

In a British lead crystal manufacturing plant, 87 people were monitored for concentrations of lead in blood and tibia in vivo, the latter reflecting cumulative exposure. The duration of exposure to lead was 10.0 ± 1.1 years. The mean concentration of lead was 481 ± 18 µg/L in blood (non-occupationally exposed, 131 ± 16 µg/L) and 17.5 ± 1.9 µg/g wet weight in tibia (non-occupationally exposed, 9.4 ± 2.1 µg/g wet weight) (Somervaille et al., 1988)

Andersson et al. (1990) collected 36 personal air samples in three Swedish glassworks, one producing heavy crystal glass and two producing semi-crystal glass. The results are presented in Table 4. The glassworks producing heavy crystal had a higher air concentration of lead especially in the foundry area, than those producing semi-crystal. Of the 28 samples from the semi-crystal glassworks, only four contained concentrations above the present Swedish threshold limit value of 50 µg/m³, whereas in the heavy crystal glassworks 7/12 samples exceeded that limit. Lead was also detected in slag from inside the blowpipes at concentrations (geometric mean) of 6.93 µg/mg slag in the heavy crystal glassworks and 0.72µg/mg slag in the semi-crystal glass industry (see Table 7, p. 359).

Table 4

Concentrations of lead (µg/m³) in the air of three Swedish glass factories, presented as geometric means (GM) with 95% confidence intervals (95% CI).

Table 7

Content of some metals (µg/mg slag) in blowpipes, presented as geometric means (GM) with 95% confidence intervals (95% CI).

Although stained-glass workers are not involved in the manufacture of glass, they may be exposed to lead during soldering Landrigan et al. (1980) measured blood lead concentrations in 12 professional stained-glass workers, in five hobbyists and in four family members of workers, in the USA. The concentrations in professional workers (mean, 207 µg/L) were higher than those of hobbyists (116 µg/L) and family members (113 µg/L)

The mean lead concentration in settled dust samples from a stained-glass workshop was 11000 ppm (mg/kg).

Baxter et al. (1985) surveyed 47 workers in the four largest stained-glass workshops in the United Kingdom. The mean blood lead concentration was 300 µg/L (range, 100–700 µg/L); nine workers had concentrations of 400 µg/L or more. The mean concentration of lead in personal air samples of five glaziers inside the workshop was 30 µg/m³ (range 10–50 µg/m³), and samples of general air all contained 10 µg/m³. All results were well below the current standard for lead in air in the United Kingdom of 150 µg/m³.

1.3.3. Exposure to other metals

Exposures in the production of hypodermic syringes from glass containing small amounts of antimony and cerium oxide were studied. Environmental concentrations of antimony, stibine (antimony hydride) and cerium were approximately 1% of the NIOSH recommended maximal concentration at the time of the study: 500 µg/m³, 0.1 ppm and 5000 µg/m³ (Burroughs & Horan, 1981).

Lüdersdorf et al. (1987) studied external and internal exposure to antimony of 109 male workers in two German glass-producing factories. Airborne, blood and urinary concentrations of antimony are presented in Table 5. The airborne concentrations in the batch area exceeded the German ‘technical exposure limit’ value of 300 µg/m³ (Deutsche Forschungsgemeinschaft, 1992), and the highest concentrations of antimony in blood and urine were found in workers in the batch area.

Table 5

Time-weighted average airborne concentrations of antimony in total dust and antimony concentrations in blood and in urine in two German glass factories.

Chrostek et al. (1980) investigated 35 crystal glassworkers working in the mix and melt area and batch house who were exposed to various compounds, including arsenic trioxide, selenium and silica flour. Personal air monitoring of eight workers revealed arsenic concentrations of 2–11 µg/m³, all of which were in excess of the NIOSH recommended standard of 2 µg/m³, although only one exceeded the US Occupational Safety and Health Administration standard of 10 µg/m³. All eight personal samples indicated exposure to selenium (0.04–7.92 µg/m³) below the US Occupational Safety and Health Administration standard of 200 µg/m3 [no NIOSH standard given]. The personal samples for respirable dust indicated exposures of 0.14–0.99 mg/m³. The quartz content of bulk dust samples was 10%. The results of clinical examinations and interviews revealed no arsenic-related health complaint or symptom and no arsenic-related skin disorder. All blood samples contained concentrations of arsenic below the detection limit of 10 µg/kg; however, blood was considered to be a less reliable specimen than urine for assessing exposure to arsenic.

Roels et al. (1982) measured total airborne arsenic in a Belgian glassware factory where arsenic trioxide was used to produce uncoloured glass, determined urinary inorganic arsenic and its metabolites and evaluated hand and mouth contamination by arsenic before and after the workshift in 10 workers. The results suggested that the high urinary arsenic concentrations found (mean, around 300 µg/g creatinine; range, 10–941 µg/g) were probably more closely related to oral intake from contaminated hands than to absorption from the lungs. Urinary arsenic excretion in control workers ranged from 7.6 to 59 µg/L.

In two German cross-sectional studies (in 1976 and 1981) to evaluate internal exposure of employees in the heavy crystal industry, urinary arsenic concentrations of 3–114 µg/g creatinine were found. In 33% (1976) and 16% (1981) of the cases, urinary arsenic elimination exceeded the upper limit of normal (25 µg/g creatinine) (Schaller et al., 1982).

In a study in the United Kingdom, two cases of subacute arsenic poisoning were reported among decorative glass workers. Although air levels in most cases were below the recommended limit of 0.2 mg/m³ (Table 6), excessive uptake seemed to have occurred in some workers (Ide & Bullough, 1988).

Table 6

Results of personal and background sampling of arsenic in air of decorative glassworks in the United Kingdom (mg/m³).

Farmer and Johnson (1990) reported exposure to airborne arsenic trioxide, used as a decolourizing agent, in a specialist glass manufacturing industry during the weighing of constituents for glass batches and during mixing of chemicals. Inorganic arsenic (As[V], As[III]) and its metabolites (mono- and dimethylarsonic acid) were determined in 18 first-void urine samples by direct hydride generation-atomic absorption spectrometry. The mean urinary arsenic concentration was 79.4 µg/g creatinine, in comparison with 4.4 µg/g creatinine for controls.

In Swedish heavy crystal and semi-crystal glass industries, Andersson et al. (1990) detected only a very small amount (< 6 µg/m³) of arsenic; nickel and manganese were not detected (< 1 µg/m³). Arsenic, nickel and manganese were detected in microgram per milligram concentrations in slag from inside blowpipes (Table 7).

Raithel et al. (1991) found very high external and internal exposure to nickel in German shops in which nickel-armoured hollow-glass moulds were produced or repaired. Hollow-glass moulds used in automatic machines are plated with nickel and nickel alloys (70–98%) by flame spraying with nickel-containing powder (75–99%); the moulds also undergo abrasive buffing and polishing with grinding disks, lathes and emery paper. The airborne nickel levels in three factories in 1981 and 1984 are summarized in Table 8. The German technical exposure limit of 500 µg/m³ for nickel and its compounds was exceeded at three of 24 measuring stations (range, 3.4–623 µg/m³). Between 1981 and 1984, dust control and working conditions were improved. Urinary nickel excretion of workers in 24 German hollow-glass factories ranged from 0.2 to 60.2 µg/g creatinine (mean, 4.5) in pre-shift samples and were 0.5–211 µg/g creatinine (median, 6.9) in post-shift samples. The highest urinary nickel concentrations were detected in flame sprayers (median, 25.3 µg/L). Lower internal exposure was found for craftsmen, i.e. polishers and chasers (median, 7.4 µg/L).

Table 8

Measurements of nickel fine dust in three German hollow-glass mould repair shops in 1981 and 1984.

1.3.4. Other exposures

Exposure may occur to polycyclic aromatic hydrocarbons in fumes generated by oil-fired furnaces and in mineral oils used for lubricating moulds. Workers were exposed by inhalation to oil aerosols during casting, where contact of oil with hot moulds at ca. 300–700 °C produces mists and fumes, in two Italian plants. In a plant producing bottles and jars for foods, moulds were lubricated manually with graphited oil; in a crystal glass plant, the moulds were lubricated automatically every few seconds with graphited oil. Typical concentrations of polycyclic aromatic hydrocarbons in the oils were 0.1–5 ppm. A number of airborne compounds (benzo[a]anthracene and chrysene) were detected in close proximity to their source during aerosol emission (i.e. during mould lubrication) at concentrations in the order of 1 µg/m³ in both plants. Such levels decreased to about 0.1 µg/m³ in personal air samples. Exposures to oil mist were in the range 0.7–2.4 mg/m³ (time-weighted average), thus complying with the generally accepted limit of 5 mg/m³ (Menichini et al., 1990).

Gunter and Thoburn (1985) measured exposures to airborne petroleum distillate in a US crystal facility, where leaded crystal was cut, grinded, polished and glued together into various art objects. The environmental concentrations were well below the NIOSH evaluation criteria, as were the concentrations of 1,1,1-trichloroethane and toluene in the breathing zone.

Interleaving materials, such as Lucor™ (50:50 mixture of Lucite® beads, a polymer of the acrylic monomer methyl methacrylate, and adipic acid) and wood flour or a combination of the two, are used in the flat glass industry to prevent window panes from adhering to each other during packing and unpacking. In a study of a flat glass industry in the USA, exposures to adipic acid were below the detection limit (2 µg per sample). Results of sampling for particulates (< 2 mg/m³) showed that all samples were below the threshold limit values of the American Conference of Governmental Industrial Hygienists and the permissible exposure limits of the US Occupational Safety and Health Administration for both total (10–15 mg/m³) and respirable (5 mg/m³) nuisance dusts (Almaguer, 1985).

2.3.1. Cancer of the urinary bladder

In a Canadian population-based study of 480 male and 152 female incident cases in 1974–76 (Howe et al., 1980), an odds ratio of 6.0 (95% CI, 0.7–276) was found for male ‘glass processors’.

A group of 303 incident cases of carcinoma (or papilloma not specified as benign) of the lower urinary tract in white males and 296 white male referent subjects were studied in Detroit (USA) in 1977–78. Patients and controls were interviewed to obtain lifetime work histories: Six case subjects and one referent subject (odds ratio, 5.9; 95% CI, 0.7–49.8) had ever been employed in glass or glass products manufacture (Silverman et al., 1983).

In a hospital-based case–control study on work-related cancer of the urinary bladder in France in 1984–87 (Cordier et al., 1993), 15 of 658 cases among men occurred in ‘glass formers and potters’ versus seven among the referents (odds ratio adjusted for smoking, 1.82; 95% CI, 0.73–4.53). [The Working Group noted that specific information on exposure was not given.]

2.3.2. Cancers of the respiratory organs

Milne et al. (1983) carried out a case–control study of 925 lung cancer deaths in Alameda County, California, USA, in 1958–62, comparing their occupations with those of 6420 deaths from other cancers in the same County, as identified from death certificates. They found a significant (p < 0.05) positive association with employment in glass, clay and stone manufacture (11 cases; odds ratio, 1.9). [The Working Group noted the limitations of using information on occupation derived solely from death certificates.]

Buiatti et al. (1985) reported a study comprising 376 incident, histologically confirmed cases of lung cancer in Florence, Italy, admitted to a regional hospital between 1981 and 1983. A total of 892 referent subjects were recruited from among patients who were admitted to the same hospital and who did not have lung cancer. Detailed work histories were compiled directly for each subject. Only one case but five referents had glass-work in their work histories [crude odds ratio, 0.5; 95% CI, 0.1–3.9].

Levin et al. (1988) studied the lifetime work histories, smoking histories and other exposure factors among 733 male Chinese with lung cancer in Shanghai in 1984–85 and among 760 male referents selected from the general population. Information was compiled by the use of a structured questionnaire administered in the subjects' homes, in hospital or at work sites. About 35 major occupational categories were examined. A deficit (odds ratio, 0.6; 95% CI, 0.3–1.5; adjusted for each occupation) of lung cancer was found among males working in ‘glass, ceramic and enamelled product manufacture’. A parallel study was performed among 672 female cases and 735 referent subjects, but small numbers in each occupational group precluded detailed analyses. The largest excess (odds ratio, 5.1; CI, 1.3–23.5) was found among female glass and glass product manufacturing workers, based on 15 cases and three referents. When 20 years of employment in the industry was applied as an inclusion criterion, the excess was more than seven fold.

2.3.3. Tumours at multiple sites

A preliminary study in three parishes with glass industries in southeastern Sweden (Wingren & Axelson, 1985) revealed an excess of deaths from stomach cancer in 1951–79 among glass-blowers (odds ratio, 6.4; 95% CI, 3.0–14.0), based on eight cases. For unspecified glass-workers [known to include glass-blowers], the odds ratio for stomach cancer was 1.5 (0.68–3.2), based on nine cases. Lung cancer was also seen in unspecified glass-workers, with an odds ratio of 2.4 (1.0–5.8), based on seven exposed cases. The same authors (Wingren & Axelson, 1987) expanded the study on mortality among workers in Swedish glass-works by selective causes of deaths, i.e. for total cancer, stomach cancer, colon cancer, lung cancer and cardiovascular deaths. Cases with these causes of death were considered in males aged 45 and over in the registers of deaths and burials in 11 parishes (including the three studied previously) in 1950–82; the eight new parishes were also analysed separately. Control subjects were taken as other causes of death among males aged 45 years and over in the same parishes. Information on exposure to various metal compounds was obtained from seven existing glass-works in the area. Stomach cancer appeared in excess for glass-blowers (odds ratio, 2.6 [95% CI, 1.4–4.9]) based on 11 exposed cases, as did colon cancer (odds ratio, 3.1 [95% CI, 1.2–7.7]), based on five exposed cases, and lung cancer (odds ratio, 2.3 [95% CI, 0.8–6.3]), based on four exposed cases. For unspecified glass-workers, the odds ratio for for stomach cancer was 1.4 [95% CI, 0.8–2.4], based on 18 exposed cases, and that for colon cancer was 1.8 [95% CI, 0.9–3.7], based on nine exposed cases; for lung cancer, the odds ratio was 1.9 [95% CI, 1.0–3.7], based on 11 exposed cases. On the basis of the information on exposure obtained from the glass-works, the authors later made an attempt to identify certain exposures as determinants of the cancer risks found (Wingren & Axelson, 1993) through further case–control evaluations in the art glass industry. The risk for stomach cancer in particular was associated with exposure to arsenic, copper, nickel, manganese and to some extent lead and chromium. For colon cancer, an increasing trend in risk was seen with increasing use of antimony and lead, the two elements being strongly correlated. For lung cancer, no obvious trend with exposure to any metal could be found.

In a case–control study based on death certificates, 9663 white and 3253 black men in Illinois (USA), aged 35–74, who had cancers of the stomach, pancreas, lung, prostate, urinary bladder or brain or non-Hodgkin's lymphomas (Mallin et al., 1989) were compared with control groups chosen by sampling randomly among non-cancer deaths from the same age groups. Occupation was coded on the basis of the 1980 census. A three-fold excess risk for brain cancer was found, based on eight cases among manufacturers of glass and glass products. [The Working Group noted the limitations of using information on occupation derived solely from death certificates.]

A case–control study conducted in the Montréal (Canada) metropolitan area (Siemiatycki, 1991) included all males aged 35–70 with a histologically confirmed diagnosis of cancer at 20 selected sites made between September 1979 and June 1985. A total of 3730 cases were interviewed. Occupational exposure to 293 substances potentially present in the work environment was assessed by a group of experts on the job description in the questionnaires. Cases of cancer at each site were compared with all other cancer cases. Age, cigarette smoking and a number of other potential confounders were controlled for in the analyses. ‘Glass dust’ was one of the exposure categories with a prevalence of 1% in the total sample and was associated with stomach cancer (5 exposed cases; odds ratio, 1.8 [90% CI, 0.7–4.8]) and lung cancer (18 exposed cases; odds ratio, 2.2 [90% CI, 1.0–5.0]). When only substantial exposure was considered, the figures were: stomach cancer, four exposed cases; odds ratio, 2.7 [90% CI, 0.8–8.6] and lung cancer, six exposed cases; odds ratio, 1.0 [90% CI, 0.3–3.6].