Concentration and size of asbestos in water supplies.

A review of the results of over 1500 asbestos analyses from U.S. water supplies suggests that the majority of water consumers are not exposed to asbestos concentrations in their drinking water over 1 x 10(6) fibers per liter. There are, however, some populations that are exposed to waterborne asbestos concentrations over 10 x 10(6) fibers per liter caused by natural erosion, mine processing wastes, waste pile erosion, corrosion of asbestos cement pipe, or disintegration of asbestos tile roofs running into cisterns. The distribution of fiber sizes in the water is dependent on the source of the fibers. The average length of chrysotile fibers found in an asbestos cement distribution system was 4 micrometers, while the average fiber length of chrysotile fibers contributed to a water supply by natural erosion was 1 micrometer.

A review of the results of over 1500 asbestos analyses from U.S. water supplies suggests that the majority ofwater consumers are not exposed to asbestos concentrations in their drinking water over 1 x 106 fibers perliter. There are, however, somepopulations that are exposed to waterborne asbestos concentrations over 10 x 106 fibers per liter caused by natural erosion, mine processing wastes, waste pile erosion, corrosion of asbestos cement pipe, or disintegration of asbestos tile roofs running into cisterns. The distribution of fiber sizes in the water is dependent on the source of the fibers. The average length of chrysotile fibers found in an asestos cement distribution system was 4 pm, while the average fiber length ofchrysotile fibers contributed to a water supply by natural erosion was 1 gm.

Fiber Concentrations in Water
Drinking water is contaminated by asbestos fibers from pollution, geologic erosion, and the disintegration of asbestos cement pipe. Since 1973-74 when asbestos was first reported to be present in potable water supplies (1-4), a number of laboratories have been analyzing for asbestos in drinking water in various cities of the United States. A review of the results of over 1500 water samples analyzed for asbestos by electron microscopy suggests that several populations of U.S. water consumers have been exposed to significant numbers of asbestos fibers in their drinking water at some time.
The waste discharge from the processing of iron ore has contributed amphibole fibers to the areas of Lake Superior which supply water for the cities of Duluth, Two Harbors and Beaver Bay, Minnesota. Concentrations as high as 600 x 106 fibers/l. have been reported for Duluth water (5). Fiber counts as high as 200 x 10fi and 92 x 106 fibers/l. have been reported for Two Harbors and Beaver Bay, respec-tively. The erosion of an old asbestos waste pile is suspected to have contaminated a water supply in Kentucky with as much as 74 x 106 fibers/I. of chrysotile asbestos. Natural erosion of asbestos bearing rock formations is considered to be the source of fibers in son e water supplies of the area around San Francisco, California and in supplies near Seattle, Washington. Concentrations of chrysotile asbestos between 1 and 100 x 106 fibers/I. have been reported for a number of supplies around San Francisco and over 100 x 106 chrysotile fibers/I. have been found consistently in the water supply of Everett, Washington. One sample of water from a distribution system in South Carolina collected after a length of asbestos cement pipe which had been attacked by corrosive water contained over 500 x 106 chrysotile fibers/l. Drinking water in other asbestos cement pipe distribution systems in Florida, Kentucky, and Pennsylvania have been shown to contain concentrations of chrysotile asbestos over 10 x 106 fibers/I. In tap water drawn from cisterns using asbestos tile roofing materials for rain collection, concentrations of chrysotile asbestos over 500 x 106 fibers/l. have been found. Table 1 summarizes the distribution of reported asbestos concentrations in the drinking water of various cities in the United States. The table is based on available results from transmission electron microscopy analyses. Because the data were re-February 1980 ported by 15 different laboratories, some using different sample preparation methods, there are some disagreements over actual values. However, Table 1 suggests that asbestos is a contaminant in a significant number of U.S. water supplies. A listing of the data available on asbestos in water supplies is presented in Table 2.
Industrial discharges of asbestos were found to range from 106 to 1012 fibers/l. during an EPAsponsored survey (7). With the exception ofthe Lake Superior situation, however, it has not been shown conclusively that any discharged asbestos fibers make their way into public drinking water.
Chrysotile, a serpentine mineral, is the most common asbestos variety found in water supplies, but some amphiboles have been identified. The amphibole crocidolite, a minor constituent of asbestos cement pipe, has been found along wth chrysotile in some waters distributed through the pipe. The amphibole fibers in Lake Superior have been determined to be primarily of the cummingtonitegrunerite series ofwhich amosite, a commercial form of asbestos, is a member. There is still some debate among mineralogists as to whether amphibole fibers found in the lake water should be called asbestos fibers or cleavage fragments. Amphiboles of the tremolite-actinolite series have been found in some water supplies of the Pacific Northwest. No fibers of the asbestos variety anthrophyllite have been reported in drinking water.
Asbestos fibers in the source of a water supply can be controlled by filtration. Treatment plants in operation in Duluth and Two Harbors, Minnesota, and pilot filtration plants in Seattle and Everett, Washington, have shown that both amphibole and chrysotile fibers can be eliminated from the water supply by coagulation and filtration.
While it is estimated that some 200,000 miles of asbestos cement pipe are in use in the United States (17), reported analyses suggest that most asbestos cement pipe does not shed significant numbers of fibers into the water. The quality of water trans- ported is known to be a critical parameter in the release of fibers from the pipe. The corrosive effect ofwater on asbestos cement pipe has been described by the aggressiveness index (Al): Al = pH + log (AH) where pH is the index of acidity or alkalinity of the water in standard pH units, A is the total alkalinity (in mg/I.) as CaCO3, and H is the calcium hardness (in mg/I.) as CaCO3. Higher values of this aggressiveness index are less corrosive than lower values. Water with an Al less than 10 is considered very aggressive to many types of pipe, while Al values greater than 12 are considered essentially nonaggressive. A statistical sampling performed by our laboratory of water supplies representative of the utilities throughout the United States suggests that 16o of the U.S. water utilities have very aggressive water, which might cause fibers to be released from asbestos cement pipe.
Even if the asbestos cement pipe is not attacked by the water some intermittent high concentrations may occur as a result of improper pipe tapping. Tapping asbestos cement pipe, that is, adding a service connection to the distribution pipe, requires that a hole be cut in the pipe. Some tapping devices allow debris from the cutting to fall into the pipe where it may remain in the water for some time depending on water flow. There are tapping devices now available that flush the debris from the pipe and thus prevent the contamination of drinking water with fibers.

Fiber Size of Asbestos in Water
Methods of sizing asbestos fibers in environmental samples have not been standardized. It is generally recognized that asbestos fibers found in water are smaller than the resolving power of the light microscope techniques (18). Little waterborne asbestos fiber size data have been determined with the scanning electron microscope because problems in resolving the very thin, small chrysotile fibrils make it difficult to use the scanning electron microscope in routine water analysis. Some water sample preparation methods used for transmission electron microscopy such as the rubout technique (3,4) deliberately destroy the particle size distribution and only allow mass concentrations to be determined. Thus, to provide fiber size distribution data on drinking water samples, only direct transfer preparation methods with transmission electron microscopy are used.
Thousands of fibers have been measured in samples of drinking water analyzed according to the methods described in the EPA Preliminary Interim Procedure for Fibrous Asbestos (19). Over 7800 waterborne asbestos fibers were measured in conjunc- BDL BDL < 1 > 10 > 10 1-10, > 10 < 1 < 10 BDL < 1 BDL 1-10, > 10 < 1 1-10 1-100 > 10 < 1 NS < 1-100 <1 High count in raw water Environmental Health Perspectives Raw water Most recent data shows NS  WY aBDL = Below detectable limits of the method (no fibers were found); NS = too few fibers were found to allow an accurate concentration value (usually NS corresponds to a count less than 0.5 x 106) fibers per liter. tion with an epidemiology study in the San Francisco from natural erosion; the sample taken after the pipe Bay Area, California. The average length and width presumably contained both the natural erosion fibers of the chrysotile fibers were 1.4 and 0.040 um, re-and some from the pipe. spectively. The lengths ranged from 0.1 to 59 um. Table 5 presents some data on aspect ratio The fibers in the drinking water of this study area (length/width) which also reflect the size differences may have come from a variety of sources including between the fibers in various drinking waters. It is natural erosion of serpentine rock, pollution from the evident even in the cases where the source is natural wastes of asbestos manufacturing and possibly cor-erosion that the vast majority of the chrysotile fibers rosion of asbestos cement pipe. exceeds a 10:1 aspect ratio. The data presented in Table 3 suggest that the fiber size distribution in the drinking water is dependent Variation in Concentration and on the source of the fibers. It is apparent that the Size Data corrosion of asbestos cement pipe when attacked by aggressive water can contribute a greater proportion In the natural system the weather plays an imporof long fibers than does the natural erosion of a tant part in varying the concentration of asbestos serpentine rock formation. The distribution of fiber fibers in water over time. Cook (5) has shown at least lengths described in Table 4 shows that fibers from a fivefold increase in an amphibole fiber concentraasbestos cement pipe systems tend to be longer than tion in drinking water as a result of a storm. The naturally occurring fibers such as are found in erosion of natural serpentine rock and of asbestos California and Washington State. Statistical analysis waste piles undoubtedly increases or decreases deof the fiber size distribution before and after pending on rainfall and stream flow. Asbestos conasbestos-cement pipe length in California showed centrations in asbestos cement pipe are known to be that the fiber set in the water before the pipe had a increased temporarily 10-or 100-fold by pipe tapping higher proportion of shorter fibers than the fiber set and are probably affected by water flow rates and after the pipe (20). The sample taken before the as-changes in water chemistry and temperature. bestos cement pipe contained fibers presumably Differences in methodology for analyzing asbestos  samples can also contribute to the variation in reported asbestos concentrations. The Nuclepore Jaffe Wick technique (21) for sample preparation has been shown to provide good interlaboratory comparison data. Laboratories using this method have reported results within a factor of two and it is gaining acceptance as the most widely used method. Differences in methodology for preparing samples for sizing can lead to differing size characterizations of the same sample. Three laboratories determined the fiber length distribution for the Union Internationale Contre le Cancer (UICC) Amosite standard reference material. The data are given in Table  6. It is evident that while Laboratories 1 and 2 found the fibers to be 40-60% under 1 ,um, Laboratory 3 found only 7% less than 1 ,um.
In environmental water sampling the water is filtered through a 0.1 micrometer pore size Nuclepore filter. A section of the filter is attached to a glass slide and a deposit of carbon evaporated onto the particulates and filter. A small section is cut and placed on an electron microscope grid. The filter is dissolved by using a modified Jaffe wick apparatus (22), leaving the particulates embedded in the carbon film on the grid. When possible, photographs are taken of random fields at 1000x and 500x magnification. Fiber lengths and diameters are measured with a 7-power measuring eyepiece on enlargements representing 3,000x and 15,000x magnification. In many water samples, however, interfering debris does not allow the fibers to be concentrated on a filter so that sev-eral fibers are present in each field of view. Often many fields of view must be searched to find one fiber. In these situations fibers are measured by aligning them with marks or circles inscribed on the fluorescent microscope screen. Replica gratings are used to determine the exact magnification at the screen's surface.
Samples which contain a number of long fibers are difficult to handle. In some cases grids with larger mesh size are used. When a fiber overlaps a grid bar, a switch from the transmission to the scanning mode allows the analyst to follow the fiber to its end.

Conclusions
Based on the available data on waterborne asbestos it is concluded that the majority of U.S. water consumers are not exposed to constant concentrations of asbestos fibers above 1 x 106 fibers per liter. In some areas, however, people are exposed to concentrations of asbestos fibers between 1 and 100 x 10fi million fibers per liter from natural erosion, pollution, or corrosion of asbestos materials such as asbestos cement pipe or roofing material.
The sizes of asbestos fibers in drinking waters differ depending on the source of the fibers. Fibers contributed by natural erosion are generally shorter than those contributed by asbestos cement pipe.
The use of a specific manufacturer's name is for identification purposes only and does not constitute endorsement by the U.S. Environmental Protection Agency. Comments on "Concentration and Size of Asbestos in Water Supplies" William E. Smith (Fairleigh-Dickinson Univ., Madison, N.J. 07940): Findings of fibers in water supplies bring up the question of whether they present any hazard to people drinking the water. To develop experimental information on this question, we maintained 600 hamsters from the age of about two months throughout their lives on drinking water with and without addition of some mineral fibers.
We found some tumors that may be related to treatment in hamsters that drank water containing fibers of naturally crystallized amosite asbestos from South Africa. No tumors attributed to treatment occurred in hamsters that drank water containing tailings from milling of taconite ore rich in cummingtonite/grunerite mineralogically related to amosite.
Since carcinogenicity of mineral fibers has been related to their dimensions by results of intrapleural injections in experimental animals, it is of interest to look at dimensions of fibers in our drinking water exposures.
In considering dimensions of fibers, I was glad to see that Dr. Millette's data did not show merely mean dimensions. Mean dimensions were not impressively different in our sample of amosite as compared to our samples of tailings as measured by electron microscopy at 2500x. However, differences in fiber length distributions were obvious, as better seen in measurements at 600x, which show that the percent of fibers longer than 10 ,um was 14% in the amosite as compared to 4% in the tailings. Among fibers measured as longer than 10 Am, 25% were longer than 20 .m in the amosite; none were longer than 20 ,jm in the tailings.
Following Dr. Millette's paper, a question was asked as to how mineral fibers were suspended in water to assure ingestion by animals. I responded as follows: To maintain suspension ofinsoluble mineral fibers in water, and thus assure their ingestion by animals drinking the water, we designed drinking fountains in which the water was put in funnels and agitated with a stream of air. We described this method in a paper last year [Am. Ind. Hyg. Assoc. J. 39: 583 (1978)].