Peter Herman herman at
Thu Aug 24 05:17:17 EST 1995

K N and P J Harris (ecoli at wrote:
: For these awkward little beggars Peter Herman is quite right.
: BUT, before you launch out onto a survey take a look at the confidence 
: intervals for MPN methods. They give a number, OK, but the +/- intervals 
: are horrific unless you want to do huge numbers of replicates per 
: dilution. There are computer programmes which allow you to get MPNs from 
: virtually any combination of dilutions and replicates.
: To get really good numbers it is worth thinking about multiwell plates 
: and some degree of automation. We had to get into this for soil 
: protozoan numbers some time ago.
: Otherwise it might be necessary to find an external examiner/assessor 
: who does not believe in statistics !

I concur with Peter Harris on the CIs but, alas, MPNs are often all you 
can do.  The 96 well plates and a 8-tip pipetter work quite well in most 
cases.  I have used them with good success for "normal" heterotrophs that 
have too high a proportion of swarmers to use dilution plates. The 96 
well plates work fine for cultures that are up in less than 10 days or 
so.  Most any elisa reader can be coerced into giving you a print-out and 
file automatically which beats the tar out of hand scoring!

The problem with 96-well plates and NH4 oxidizers is the long incubation 
times.  At home in New Mexico, we run into drying problems.  The little 
so and so's are aerobes which like a reasonable oxygen tension, so you 
can't really seal them to prevent evaporation.  Perhaps I can get away 
with it on sabbitical here in Sweden where it appears our nice dry summer 
has just ended.

As an aside on the CI problem.  In a number of test runs, we found that 
we were better off running more 3-tube MPNs then fewer 5-tube MPNs when 
dealing with environmental samples (I can't say for culture isolates 
where the inherent variability in sampling from "the pot" is lower than 
in the soil).  You run into a physical barrier of how many tubes you can 
process in a day and how many you can afford to have tied up in the 
pantry incubating for a month.  I strongly recommend that anyone run a 
pilot study to see if (for example) 3, 3-tube MPNs are better than 2 
5-tube runs or 5, 3-tube runs are better than 3, 5-tube runs.  Your 
answer will result from the marriage of the inherent MPN CI problems and 
the variability of your sampled source.  In our case, the high sample 
source variability swamps the MPN variability

Well - there is a long ramble for you!


R. Peter Herman				email	Peter.Herman at
Sveriges Lantbruksuniversitet		Phone:	+46 18 67 12 20
Inst. f. Markvetenskap			Fax:	+46 18 67 27 95
S750 07 Uppsala, Sweden		

More information about the Microbio mailing list