or in tubes of diagnostic media. Mistakes and errors can be minimised by proper training and supervision of samplers and analysts in their duties. In addition, laboratory quality controls and the careful application of standard procedures for analysis are essential. Standard procedures should be written correctly and copies should always be available for reference in the laboratory.
The multiple tube method is very sensitive for the detection of a small number of indicator organisms, but the MPN is not a precise value. Confidence intervals, i.e. most probable range (MPR), are often published with the MPN and are meant to indicate the imprecision of the method (Tillett, 1995). However, it should be stated clearly that the range applies to the sample and not to the water source (PHLS, 1994).
For thermotolerant coliforms and faecal streptococci membrane filters with 20-60 typical colonies are recommended for counting, with the provision that filters with no more than 200 colonies of all types should be considered; if the counts of colonies on the membranes are all below the minimum recommended, they should be totalled (APHA/AWWA/WPCF, 1995). Counting colonies on all filters has been shown to improve precision (Gameson, 1983) and is the only method specified in other standard procedures (ISO, 1988, 1990b). The count, in cfu per 100 ml, becomes the sum of all colonies counted multiplied by 100 ml and divided by the total volume (in ml) of water filtered.
If confirmation tests have been applied to a number of typical and atypical colonies, the initial count should be adjusted by multiplying it by the percentage of verified colonies. This percentage will be calculated by dividing the number of verified colonies by the total number of colonies subject to verification and then by multiplying the result by 100 (PHLS, 1994; WHO/UNEP, 1994a; APHA/AWWA/WPCF, 1995). This procedure has a considerable effect in reducing the precision of the count, depending on the total number of presumptive colonies selected and the fraction confirming (PHLS, 1994). Nevertheless, all colonies should be selected if there are ten or fewer colonies on a membrane. New proposed ISO methods try to overcome the imprecision by proposing verification methods for all the colonies grown on the filter (Table 8.7).
If the total number of bacteria colonies, including the specific target colonies, exceed 200 per membrane or are not distinctive enough to enable counting, results should be reported as “too numerous to count” (APHA/AWWA/WPCF, 1995) or “count too high to be estimated at the dilution employed” (PHLS, 1994). A new sample should be requested immediately if possible and more appropriate volumes should be selected for filtration (Table 8.2). The resultant data however, represent the results from a different sample; nevertheless the data may help to investigate the event. If this approach is not possible, it is preferable to try to count a sector of the original filter, before the information is lost thereby estimating the total number of colonies (even though these counts may lack precision). This technique is facilitated by the grid printed on the membranes. The details of the estimate should be recorded, for example “Count in n squares = x; diameter of filtration circle = d squares; estimated count = πd2x/4n colonies”. If 135 colonies were counted in 10 squares and the diameter of the filtration area was 11.5 squares, the count would be estimated to two significant figures as 3.142 × 11.52 × 135/40 = 1,400 colonies.