The use of enterococcus and coliform in characterizing bathing-beach waters.

Introduction

Section 6-2.15 of the New York State Sanitary Code (1) establishes the standards for water quality and monitoring of New York's bathing beaches. The bacteriological standard currently in use refers to the numbers of either total or fecal-coliform bacteria, the choice being left to the discretion of the local health departments. Although no evidence of undue illness resulting from bathing in properly monitored state waters has been found (indeed, there is some indication that the coliform standard may be too restrictive in certain areas), studies have been performed that cast doubt on whether coliform bacteria can provide a true indication of bathing-water quality relative to public health (2-5).

Dufour (2) recounts the history of the use of total- and fecal-coliform bacteria for classifying bathing waters and provides a discussion of works sponsored by the U.S. Environmental Protection Agency (U.S. EPA), which examine the relationship between water quality and swimming-associated illness (3,4,5). The studies indicated that at the bathing areas investigated, no relationship existed between total or fecal coliforms and rates of gastrointestinal (GI) symptoms (the most common swimming-related malady). A significant correlation was found between enterococci and GI symptoms, however, and a slightly lower correlation between E. coli and GI symptoms was found in marine waters (3). In freshwater, the correlation coefficient for E. coli was higher, though not significantly so, than that for enterococcus (2,5). More recent studies in the United Kingdom (6) have noted a statistical relationship between gastroenteritis from sea bathing and fecal streptococci measured at chest depth. However, the organisms defined as fecal streptococci in the British publication are essentially the same organisms defined as enterococci in the U.S. literature.

Although the mathematical relationship between enterococci density and gastroenteritis derived by Cabelli (3) has been questioned (7), the studies resulted in the U.S. EPA recommendation to use enterococcus as an indicator of bathing-water quality, with E. coli being an option for freshwater (8). This recommendation, along with that of the Natural Resources Defense Council (9) (a national environmental advocacy group) for a national bathing-water standard, was the impetus for this investigation.

The present New York State regulatory program is based on current and historical coliform data, sanitary surveys, rainfall, and local health departments' expertise and firsthand knowledge of their bathing environments. As there has been no indication that the present program is not sufficiently protective of the public health, officials are reluctant to modify the program without first investigating the potential effect of such modification on beach closure days - that is, it would be counterproductive if the number of closure days were increased without a concomitant increase in protection of the public health. Additionally, the question of the universality, or lack thereof, of any monitoring program must be addressed. Cabelli (3) states quite clearly of the enterococcus-gastroenteritis relationship that this criterion is directed against potential human health effects consequent to the pollution of marine recreational waters with human fecal wastes, notably municipal sewage. It is a generalization which may not always hold true. (Emphasis added)

TABLE 1

Correlation Matrix

                                All Samples (n = 791)

Total Coliform    Fecal Coliform    Fecal Coliform    Enterococcus

Total Coliform         1.00               -                -
Fecal Coliform          .90             1.00               -
Enterococcus            .70              .58              1.00

[TABULAR DATA FOR TABLE 2 OMITTED]

Cabelli's point holds even for the coliform standard currently used, in which the presence of high numbers of fecal-coliform organisms may indicate the presence of human sewage (with a direct public health relationship) or of animal wastes (that may have less public health significance). Beach closures due to contravention of coliform standards in areas lacking any municipal sewage source are likely to be conservative. Thus, the desire for a national standard, while laudable, must be approached with caution.

During the 1994 bathing season, the Suffolk County Department of Health Services (SCDHS) analyzed bathing-beach water samples for members of the enterococcus group, in addition to the routinely analyzed total- and fecal-coliform bacteria, in an effort to determine the utility of each group for the characterization of bathing-beach waters.

Of interest were the following questions:

* What effect would using the U.S. EPA beach closure guidance instead of the coliform standard have on beach operations within Suffolk County - that is, would the frequency of standard violation increase, decrease, or remain the same?

* What would be the costs, including those for laboratory staff and related expenses, of changing from the present standard to an enterococcus standard?

Most important, but perhaps not easily answered in light of Cabelli's previously noted caveat:

* Will an enterococcus standard provide better protection of the public health than the present standard without being unduly restrictive?

Area of Study

Areas sampled include freshwater (lakes and ponds) and seawater beaches in Suffolk County [ILLUSTRATION FOR FIGURE 1 OMITTED]. Seawater beaches ranged from extremely sheltered and semi-enclosed bay beaches to open-water beaches on Long Island Sound and the Atlantic Ocean.

Methods

In total, 1,626 water samples were collected by SCDHS staff from 177 bathing beaches between June 1 and September 15, 1994. Samples were collected from just below the water's surface in clean, presterilized (auto-claved) polypropylene bottles with 125-mL-wide mouths and placed immediately into an ice chest containing sufficient ice to maintain sample temperatures below 10 [degrees] C, and were delivered to the Environmental Health Laboratory (EHL) within six hours of collection. A temperature control tube was included with the samples. Sampling frequency depended on the expectation, based on experience and historical data, of elevated coliform values. Freshwater beaches and beaches within semi-enclosed harbors and bays were sampled at least five times per month. Beaches on more open bodies of water (Long Island Sound, Peconic Estuary) were sampled less frequently (one to four times per month, depending on the location), and ocean beaches were sampled least often.

All samples were analyzed for total and fecal coliforms by the Multiple Tube Fermentation Technique (10). The mE Membrane Filter method (10) was used for the analysis of the enterococcus group, which began with samples collected on July 19, 1994. The mE medium is the most specific for enterococci (11-13) and was used by Cabelli et al. (4,14) in their studies of recreational water quality. It is also the procedure used by the neighboring state of Connecticut for evaluation of its beaches. A comparison of the mE and multiple tube procedures for the determination of enterococcus (10) was made by the Suffolk County EHL on samples collected from 17 different beaches during the 1995 bathing season. The various methods for the enumeration of enterococci have been discussed by U.S. EPA (15). The Suffolk County EHL is certified by the New York State Environmental Laboratory Approval Program.

Data Handling

Dilutions used for the total- and fecal-coliform multiple-tube procedures provided most probable number (MPN) values ranging from less than or equal to 20 to greater than or equal to 16,000. When either of the boundary values were reached, that value was used. All coliform values refer to the most probable number per 100 mL (MPN/100 mL).

Results

In all, 791 water samples collected from 143 sites from mid-July through mid- September were analyzed for enterococcus and for total and fecal coliforms. The correlation coefficients for the paired values (Table 1) were 0.70 and 0.58 respectively for enterococcus versus total coliform and for enterococcus versus fecal coliform. The correlation between total and fecal coliform was 0.90.

From mid-July through mid-September, the total coliform value exceeded the state standard of 5,000 per 100 mL on 10 separate occasions involving nine beaches, and fecal coliform exceeded the state standard of 1,000 per 100 mL on 26 separate occasions involving 22 beaches. In four of these cases, the enterococcus value (reported as enterococci per 100 mL) was less than the U.S. EPA-recommended standard of 104.

TABLE 3

Number of Beaches and Minimum Number of Closure Days(1) that Would
Have Occurred with Use of Various Standards(2)

                                Number of              Number of
Instantaneous Value         Beaches Affected         Closure Days

Total Coli (5,000)                  9                     12
Fecal Coli (1,000)                 22                     28
Enterococcus (104, EPA)            44                     64
Enterococcus (61, Conn.)           58                    101

30-Day Log Average

Total Coli (2,400)                  0                      0
Fecal Coli (400)                    4                      9
Fecal Coli (200)                    8                     24
Enterococcus (35, EPA)             10                     34
Enterococcus (33, Conn.)           11                     38

1 Does not include time necessary for subsequent sample analysis,
time between sample collection, or time required for the log average
to drop below the standard value.

2 Actual closures resulted from fecal-coliform log averages in
excess of 400 per 100 mL and included Scudder Park in mid-July (for
the remainder of the season, calculated as 51 days), and Huntington
Beach Community Association (HBCA)for 21 days. HBCA, as a homeowners
association, is not subject to regulation.

Note: Two beaches exceeded a fecal-coliform log average of 400 at
the end of the bathing season and are not included in the closure
statistics.

On 64 separate occasions involving 44 beaches, the enterococcus value was greater than the U.S. EPA-recommended standard of 104. In 43 of these cases, the fecal-coliform value was less than 1,000, and in 51 cases, total coliform was less than 5,000 per mL.

The enterococcus count was greater than 61 per mL (the Connecticut state standard) on 101 separate occasions involving 58 different beaches; at the same time, fecal coliform was less than 1,000 per 100 mL in 78 cases, and total coliform was less than 5,000 in 89 cases.

Table 2 provides a summary of the data collected, giving a mean of the log average values, and the median values for beaches within specific groupings: Freshwater, North Shore Harbors, Great South Bay, Peconic Bay, Gardiners Bay-Block Island Sound, Long Island Sound, and the Atlantic Ocean. Because of insufficient information, data from Shinnecock Bay, Moriches Bay, and Lake Montauk are not included in the table or in the following analysis.

Mean total-coliform log averages were greatest at Great South Bay beaches followed closely by freshwater beaches and North Shore Harbor beaches [ILLUSTRATION FOR FIGURE 2 OMITTED]. Beaches in Long Island Sound, Gardiners Bay-Block Island Sound, Peconic Bay, and the Atlantic Ocean had significantly lower means. Mean fecal-coliform log averages [ILLUSTRATION FOR FIGURE 3 OMITTED] were somewhat different, with the North Shore Harbors and Great South Bay exhibiting the highest values and values for the rest of the beaches being considerably lower.

Enterococcus mean log averages, although much lower, correlated well with both total- and fecal-coliform mean log averages (r = 0.89 and 0.96 respectively). Medians, which, because of the nature of the variables involved, may be more representative of actual conditions within a given area, are similarly related, having r values of 0.96 and 0.99 respectively.

Comparison of mE and Multiple Tube Methods

The enterococcus values of 17 beach samples collected by the SCDHS and analyzed by EHL using both methods ranged from five to 2,500 organisms per 100 mL with a mean of 199 for the mE procedure, and from two to 1,600 organisms per 100 mL with a mean of 171 for the multiple tube procedure. The Pearson correlation coefficient was 0.99.

Cost Analysis

Although no change in the cost of sample collection would be expected, it is estimated that laboratory analysis of enterococcus by membrane filtration would cost about one-third less than MPN coliform analysis.

Discussion

A comparison of the data allows us to address the first question posed in the introduction: What effect would using the U.S. EPA beach closure guidance instead of the coliform standard have on beach operations within Suffolk County - that is, would the frequency of standard violation increase, decrease, or remain the same?

It should be noted that calculation of the actual number of closure days under each standard is artificial, because most health jurisdictions do not base beach closures solely on a single violation of a bacteriological standard. Other factors considered include a comparison of recent and historical data trends; results of sanitary surveys; the identification and sampling of possible contributing sources of pollution; and, when applicable, the effects of rainfall and tide.

Moreover, when a beach is closed because of unacceptable bacteriological quality, the beach is resampled immediately thereafter and on a more frequent schedule for the rest of the season. The change in sampling frequency and the time required for additional analyses further confound the problem of determining closure frequency and time period.

Nonetheless, for comparative purposes, closure days have been calculated and are based on violations of the following standards, with the exception that not all beaches were sampled five or more times within a 30-day period:

* Coliform standard: Subpart 6-2.15 of the New York State Sanitary Code (1) states that

(1) The total number of organisms of the coliform group shall not exceed a logarithmic mean of 2400/100 mL for a series of five or more samples in any 30-day period, nor shall 20 percent of total samples during the period exceed 5000/100 mL. When the above described standards are exceeded, the permit-issuing official shall cause an investigation to be made to determine and eliminate the source or sources of pollution; or

(2) The fecal coliform density from a series of five or more samples in any 30-day period shall not exceed a logarithmic mean of 200 per 100 mL. When fecal coliform density of any sample exceeds 1,000 per 100 mL, consideration shall be given to closing the beach, and daily samples shall immediately be collected and analyzed for fecal coliform for at least two consecutive days.

* Enterococcus standard: U.S. EPA criteria recommend that

based on a statistically significant number of samples (generally not less than five samples equally spaced over a 30-day period), the geometric mean of the enterococci densities should not exceed 35 per 100 mL nor should any single sample exceed 104 per 100 mL (8).

The state of Connecticut uses an instantaneous standard of 61 and a 30-day log average standard of 33.

If an instantaneous total-coliform standard of 5,000 per 100 mL had been used for beach closure for the period of June 1 through September 15, 1994, nine beaches would have been closed - some for more than one day - for a total of 12 days. Using an instantaneous fecal-coliform standard of 1,000 per 100 mL would have resulted in the closure of 22 different beaches for a total of 28 days.

For the period beginning with the first enterococcus sampling (July 19), 44 beaches would have been closed for a total of 64 days if the U.S. EPA-recommended instantaneous standard for enterococcus of 104 per 100 mL had been used. The fecal-coliform standard was exceeded on only 19 of those days. If the Connecticut instantaneous standard of 61 per 100 mL had been used, 58 beaches would have been closed for a total of 101 days, with the fecal standard having been exceeded on only 23 occasions.

For the period beginning with the first enterococcus log average computation (August 1), the total-coliform log average standard of 2,400 was never exceeded. The fecal-coliform log average standard of 200 was reached or exceeded at eight beaches on 24 separate occasions. Three beaches accounted for 10 of those occasions. Although SCDHS officially uses the total-coliform standard to regulate beach operations, it also requests beach closure if the fecal-coliform log average of at least five samples within a 30-day period reaches or exceeds an MPN of 400 per 100 mL. In practice, this has usually occurred before contravention by the total-coliform standard. During this investigation, the 30-day fecal-coliform log average was greater than or equal to 400 per 100 mL at four beaches on nine separate occasions, with one beach (Scudder Park) accounting for five occasions. Thus, the use of a fecal-coliform 30-day log average value of 400 per 100 mL would have resulted in the closure of four beaches. Only two beaches were actually closed by the department, because the other two did not exceed 400 per 100 mL until the end of the 1994 bathing season.

Use of the U.S. EPA-recommended enterococcus 30-day log average standard of 35 per 100 mL would have resulted in 10 beaches reaching or exceeding the standard on 34 separate occasions. Connecticut's recommended standard of 33 organisms per 100 mL was equaled or exceeded at 11 beaches for a total of 38 occasions. Scudder Park accounted for eight occasions under both standards.

The beach closure days that would have occurred under the use of the various standards are summarized in Table 3 and Figure 4. Actual closures, resulting from the 30-day fecal-coliform log averages exceeding 400 organisms per 100 mL, included Scudder Park, which was closed in mid-July for the remainder of the season (calculated as 51 days), and the Huntington Beach Community Association, which was closed for 21 days.

As can be seen from Figures 2 and 3, the beaches exhibiting the highest bacterial counts, coliform or enterococcus, are found in areas where water movement and/or tidal flushing is minimal - that is, in freshwater lakes and ponds and in semi-enclosed harbors and bays. There is sufficient variation to indicate that the topography and use of the surrounding watershed, both immediate and distant, are relevant. For instance, a single sampling of Kalers Pond, a freshwater pond at which a beach was formerly located, resulted in the finding of a total-coliform value greater than or equal to 16,000 per 100 mL and a fecal-coliform value greater than or equal to 5,000 per 100 mL. Data and field observations have indicated the major source of pollution to be the large numbers of waterfowl that frequent the area (SCDHS, unpub.).

Scudder Park, where bacterial values have historically been found to be elevated, is at the very head of Northport Harbor, where water movement and flushing are extremely limited. Pollutants that enter the waters in that area are therefore likely to remain. Confounding this problem is the fact that the Scudder Park beach is also near the effluent outfall of the Northport sewage treatment plant. However, coincident sampling at the effluent site and the beach site suggests that effluent from the sewage treatment plant is generally not a problem in terms of bacterial contamination. Sampling at this beach (along with others that experience has shown to be suspect) begins before the bathing season, and the beach often is not allowed to open or is closed shortly after the season begins.

Other problematic beaches include the Huntington Beach Community Association in Centerport Harbor and the Goldstar Battalion Park beach in Huntington Harbor. Both beaches lie near storm drains that carry runoff from the surrounding hilly land. It is not surprising that they exhibit elevated bacterial values after rainfall events.

Rainfall-Related Water Quality

The bacterial water quality of the beaches in the Northport-Huntington Harbor complex appear to be quite responsive to rainfall events. The increase in the numbers of surface-water bacteria caused by rainfall and the consequent runoff has been amply demonstrated by the Long Island Regional Planning Board (16). Similar increases at beaches within Great South Bay have been attributed to stream discharge and surface runoff (17).

In tidal waters, the degree and number of tidal exchanges are extremely important in the reduction of bacterial populations introduced from the surrounding area. Because of the demonstrated relationship between rainfall, tides, and bacterial numbers, it appears that it might be possible to control beach use on the basis of rainfall data, much as the Interstate Shellfish Sanitation Conference (18) controls the designation of shellfish harvesting areas. Rainfall models are, in fact, employed by some municipalities for regulating beaches.

Because many beaches are dispersed throughout Suffolk County, collection of the information required to prepare site-specific rainfall-tidal models on which closings and subsequent reopenings can be based has not yet been possible. The ability to control beach operations based on such real-time data would be of greater public health significance than the present methodology, whatever the indicators used. Because of the limitations of culture techniques currently used in determining bacterial population densities, a significant amount of time passes before closure decisions based on those densities can be made, and a beach is likely to remain open during a period of standard contravention. Closure, in fact, may not take place until two to three days after the peak bacterial count occurs, by which time the water quality might be excellent. Additionally, limited sampling undoubtedly biases the characterization of a beach, as samples may not include particularly wet or dry weather. This problem may be mitigated by increased random sampling and a historical database sufficient to represent all weather conditions and various tidal stages.

Conclusions and Recommendations

In 1994, the use of the enterococcus standard rather than the coliform standard in Suffolk County would have resulted in an increased rate of beach closures that could not be demonstrated to have any relationship to public health protection. Although this study was not designed to assess the health effects of bathing at county beaches, it should be noted that no indication of enteric or other illnesses possibly associated with bathing was received from either the general public or local hospital and health care providers. At this time, no clear advantage is seen in changing to an enterococcus standard. Indeed, the availability of a large coliform database (over 20 years in Suffolk County), which allows for the determination of water quality trends, would seem to support continued use of coliform standards until a clearly better method of regulating beach use becomes available. The continuous long term database also allows for temporal and spatial comparisons and the identification of new pollution sources, and it can be helpful in the construction of predictive closure models.

U.S. EPA-sponsored research and research at beaches in the United Kingdom (6) adjudge enterococcus a better indicator than coliform for predicting the likelihood of gastroenteritis. More recent research in Santa Monica Bay has shown a relationship between the total to fecal-coliform ratio with health effects, and enterococcus with health effects in the vicinity of storm drains carrying "heavy urban runoff" (19).

Both the U.S. EPA studies and the Santa Monica Bay study dealt with waters subject to point-source pollution, the former from large sewage treatment plants and the latter from stormwater and unidentified non-stormwater discharge. Cabelli (3) stated that the enterococcus-gastroenteritis relationship identified in his study was a generalization that may not always hold true and that "such relationships appear to hold for waters receiving discharges from relatively large municipal sewage treatment facilities."

Kay et al. (6) demonstrated a mathematical relationship between gastroenteritis and "fecal streptococcus" which, as determined by the methodology of the studies (20) appears to be equivalent to enterococcus as defined by the American Public Health Association (10). The proximity of the beaches sampled to sewage treatment plants or other point sources was not noted, however.

The type of pollution likely to affect beaches in Suffolk County does not come from large sewage treatment plants or storm drains carrying "heavy urban runoff," but rather from nonpoint sources. Additional research is needed to evaluate the relationship between indicator organisms and public health effects in waters subject to nonpoint source inputs.

As no public health problem has been evidenced in Suffolk County, it would be unnecessary and counterproductive to use a standard that would result in increased beach closures - and to cause the social and economic impacts of such closures - without a clear public health benefit. Further, regardless of the indicator used, by the time the laboratory results are available they may be of only historical importance in that they describe the water as it was several days before, but not at the time of closure. While these data allow the development of a beach "history," they do not provide for immediate action.

Because of the relationship between surface-water bacterial load and stormwater runoff at many beaches, and the ability to use this relationship for real-time management of bathing beaches, development and use of a rainfall model for each beach area clearly seems to be the best approach to protecting the public health. The development of such predictive rainfall-tidal models, which can be calibrated and verified by routine sampling, is recommended.

Corresponding Author: Robert Nuzzi, Suffolk County Department of Health Services, County Center, Riverhead, N.Y. 11901.

Acknowledgements: The authors wish to acknowledge the assistance of Carl Vogel and Alice Peltze of the SCDHS Environmental Health Laboratory. Thanks also are due to David Dziewulski of the New York State Health Department for reviewing the manuscript and for comments regarding the statistical nature of the MPN procedure, and to Frank Guido of the Westchester County Health Dept. for assistance in manuscript preparation and review.

REFERENCES

1. State of New York Official Compilation of Codes, Rules and Regulations, Title 10 Health, Chapter 1, State Sanitary Code, Subpart 6-2 Bathing Beaches (1995), Albany, N.Y.:Dept. of State.

2. Dufour, A.P. (1984), "Bacterial Indicators of Recreational Water Quality," Can. J. Public Health, 75(1):49-56.

3. Cabelli, V.J, (1983), Health Effects Criteria for Marine Recreational Waters, EPA- 600/1-80-031, Washington, D.C.:U.S. Environmental Protection Agency.

4. Cabelli, V.J., A.P. Dufour, M.A. Levin, and L.J. McCabe (1982), "Swimming-Associated Gastroenteritis and Water Quality," Am. J. Epidemiol., 115(4):606-616.

5. Dufour, A.P. (1984), Health Effects Criteria for Fresh Recreational Waters, EPA-600/1- 84-004, Washington, D.C.:U.S. Environmental Protection Agency.

6. Kay, D., J.M. Fleisher, A.F. Godfree, F. Jones, R.L. Salmon, R, Shore, M.D. Wyer, and R. Zelenauch-Jacquotte (1994), "Predicting Likelihood of Gastroenteritis from Sea Bathing: Results from Randomised Exposure," The Lancet, 344(8927):905-909

7. Fleisher, J.M. (1991), "A Reanalysis of Data Supporting U.S. Federal Bacteriological Water Quality Criteria Governing Marine Recreational Waters," Research J. Water Pollut. Control Fed., 63(3):259-265.

8. U.S. Environmental Protection Agency (1986), Ambient Water Quality Criteria for Bacteria - 1986, EPA 44015-84-002, Office of Regulations and Standards, Washington, D.C.:U.S. Government Printing Office.

9. Testing the Waters: A Study of Beach Closings in Ten Coastal States (1991), New York, N.Y.:Natural Resources Defense Council.

10. Standard Methods for the Examination of Water and Wastewater (1992), 18th ed:, Washington, D.C.:American Public Health Association.

11. Pagel, J.E. and G.M. Hardy (1980), "Comparison of Selected Media for the Enumeration and Identification of Fecal Streptococci from Natural Waters," Can. J. Microbiol., 26(11):1320-1327.

12. Stanfield, G., T.E. Irving, and J.A. Robinson (1978), "Isolation of Faecal Streptococci from Sewage," Technical Report TR83, Stevenage, England:Water Research Centre.

13. Dutka, B.J., and K.K. Kwan (1978), "Comparison of Eight Media-Procedures for Recovering Faecal Streptococci from Water Under Winter Conditions," J. Appl. Bacteriol., 45(3):333-340.

14. Cabelli, V.J., A.P. Dufour, M.A. Levin, and L.J. McCabe (1983), "A Marine Recreational Water Quality Criterion Consistent with Indicator Concepts and Risk Analysis," J. Water Pollut. Control Fed., 55 (10):13061314.

15. U.S. Environmental Protection Agency (1986), Waterborne Diseases in the United States, ed. G.F. Craun, Fla.: CRC Press.

16. The Long Island Segment of the Nationwide Urban Runoff Program, (1982), Hauppauge, N.Y.:Long Island Regional Planning Board.

17. Dennison, W.C., L.E. Koppelman, and R. Nuzzi (1991), "Water Quality," The Great South Bay, ed., J.R. Schubel, T.M. Bell, and H.H. Carter, Albany, N.Y.:State University of New York Press.

18. Smoley, C.K. (1993), Sanitation of Shellfish Growing Areas, National Shellfish Sanitation Program Manual of Operations, Parts I and II, Fla.:CRC Press.

19. Haile, R.W., J. Alamillo, K. Barrett, R. Cressey, J. Dermond, C. Ervin, A. Glasser, N. Harawa, P. Harmon, J. Harper, C. McGee, R.C. Millikan, M. Nides, and J.S. Witte (1996), "An Epidemiological Study of Possible Adverse Health Effects of Swimming in Santa Monica Bay," Final report, Santa Monica, Calif.:Santa Monica Bay Restoration Project.

20. The Bacterial Examination of Drinking Water Supplies, (1969), Report on Public Health and Medical Subjects No. 71, London, England: HM Stationary Office.