Tel.+49 30 838 62550 Fax.+49 30 838 46029
The accuracy of two commonly used colony count methods was determined in model experiments with Escherichia (E.) coli pure cultures. The techniques used were the depth culture (pour plate method) and the surface technique (streak plate method). Furthermore the effect of the culture medium (BHI-broth versus UHT-milk) on the growing rate and other interactions of the micro-organisms was to be examined. An additional aspect deals with the relation between the date of counting and the rate ofascertainable colonies. 16 model experiments each with Brain-Heart-Infusion (BHIbroth) and ultra high temperature sterilized (UHT-) milk with pour plating method and 9 examinations each with streak technique were conducted. Every dilution series for one model experiment comprised of 7 dilution stages within the ratio 5:1 in a counting range of 0 to about maximal 1.200 colonies. For each dilution stage 25 parallel plates were established andevaluated.The results can be summarized as follows:
1. The linearity of the dilution stages is maintained throughout the counting range of high to relative low colony numbers. Only with very strong dilutions the continuous formes become irregular whereas the streak technique performed somewhat more advantageous. However, even until about 1.200 colonies/plate no deviation of the linearity is proved, despite the danger of overcrowding andoverlapping.
2. The standard deviation of the y-values around theregression line (sex) is almost of the same quality with both streak technique and pour plate method/BHI broth. A higher standard error of the estimations exists with pourplate method/UHT-milk.
3. The real proportionality of 5:1 is never exactly maintained, however the tendency exists in all experiments to fall below the given dilution factor. While in almost half of the experiments with the pour method the ideal slope of -1 is included in the 95% confidence interval, with the streak method it can only be seen in 2 of 18 experiments.
4. Ideally, which means that except the sampling error no more components of error have an influence on the plate count, the relations s2/x for the 25 parallel plates on one dilution stage follows the model of the POISSON-distribution and takes the value of 1. The pour plate method reaches an agreement with this concept in the range of 1 to about 5 x 10<pub>2</pub> colonies per plate, whereas an overdispersion is found in extremely low or high colony counts. The streak technique shows a higher variance even with colony counts from about 102 . Within the lower colony count range an almost symmetrical variation exists around the ideal value with a tendency to fall below. These results can be confirmed in the estimation of dispersion indices as well as in the boxand whisker plot"s graphics.
5. The skewness of the frequency polygon of POISSON-distributed stochastic numbers turns with increasing mean value from the positive skewed curve into a symmetric one. The same is found with the skewness calculated for the 25 parallel plates per dilution series. With the streak technique and pour plate method (BHI-broth) colony counts of more then 100 even turn into a negative skewed curve which could be explained as an impression of overdispersion.
6. Since the true amount of organisms in the samples emained unknown it is only possible to give relative statements of the accuracy of the mean from the comparison of the four experiments which were not connected with each other. It can be seen that the growing rates of the E. coli organisms in BHI broth exceed the rates in UHT milk. An interaction with the nutrient medium occured as far as the pour plate procedure with the UHT milk achieved higher colony counts while the recovery with BHI broth with streak and pour plate procedure is the same.
7. As far as the time of determination for the cultural plate count of E. coli pure cultures is concerned, neither a prolonged incubation time (72 h instead of 48 h) nor an additional storage in a refrigerator has any effects.8. For the enumeration of pure cultures of E. coli with the pour plate procedure a larger counting range then usually can be used since for colony numbers from 1 up to about 5 x 10<pub>2</pub> the variance almost fits the arithmetic mean. With the streak technique a higher variance is already found with colony numbers at 10<pub>2</pub>, but the linearity of the dilution series is kept even above these limits.
Concerning the linearity and proportionality of dilution series, mean variation of the results of paralles plates of one dilution stage as well as the recovery the KOCH"s plate method in the model experiment is proved to be superior to the streak technique in most points. The advantages in precision of analysis are payed for with a higher amount of labour and material investment. Therefore in spite of less accuracy the surface streak procedure should find its place in routine work as acceptable compromise between accuracy and expenditure.