If you look into the raw data for the IAEA C-14 intercomparison materials 
from 1990, you will find for IAEA-C3 cellulose a mean delta13C value of 
 -24.91 +- 0.56 permille vs.PDB (n=51,  here only three obvious outliers 
excluded with no further data treatment, 1sigma standard deviation).
This corresponds to -24.91+-0.08 permille (standard error of the mean) which 
should be compared with -24.64 +- 0.13 permille (vs.???)(n=5) as reported by 
Frank Pawellek.
Data were taken from the 1991 IAEA Report on the "Consultants' group meeting 
on C-14 reference materials for radiocarbon laboratories" (final results 
also published in Radiocarbon, 34, 506-519 by Rozanski et al).
As Frank pointed out correctly, IAEA-C3 cellulose should to be used as C-14 
intercomparison material and is distributed in units of 50g; reported 
delta13C values from 69 C-14 laboratories showed a relatively large scatter. 
Isotopic inhomogeneities for much smaller fractions than 50g are quite 
possible (about 1 cubic metre (!) of cellulose was prepared and homogenized 
for C-14 measurements) . I think, the existence of isotope inhomogeneities 
for small sub-samples was shown by the recent measurements.

In order to prepare a cellulose reference material for delta13C 
measurements, a considerably amount (2-3 kg) of this cellulose could be 
milled down to fine powder, homogenized and checked for isotopic 
homogeneity.
Laboratories which would be interested in such a cellulose reference 
material should send a short e-mail message to me.
Best regards,
                        Manfred Groening
++++++++++++++++++++++++++++++++++
Manfred Groening
Isotope Hydrology Section --- International Atomic Energy Agency
P.O.Box 100   A-1400 Vienna   AUSTRIA
Phone: ++43-1-206021740    Fax:  ++43-1-20607
e-mail: [log in to unmask]
NEW e-mail (valid from 17.March1997 on): [log in to unmask]
+++++++++++++++++++++++++++++++++

>----------
>From: isogeochem
>To: isogeochem
>Subject: IAEA-C3 cellulose standard: results of an inter-lab comparison
>Date: Friday, March 07, 1997 4:08PM
>
>To everybody dealing with cellulose!
>
>I had suggested a little ring test some months ago using a IAEA
>standard cellulose. Here are the results.
>
>The background:
>Since I am using a combustion technique at a rather low temperature
>for my cellulose samples, I could not use graphite standards like most
>other labs do - the graphite simply won't react. Therefore I started
>to use IAEA cellulose for internal check of my precision.
>
>The IAEA informed me that their C3 standard was likely to be
>inhomogeneous on a mg-scale  for 13 C and was not published for
>such standardization purposes, but to the best of my knowledge
>that had never been properly checked. This was for the reason of
>the proposed test.
>
>I had sent strips of C3 cellulose out to five laboratories that had
>kindly agreed to take part in the test, four of which have provided
>results so far. A summary of the results and methods used is
>given below:
>
>METHODS
>
>lab. 1: +CuO offline, 850 C, 2hrs
>lab. 2: +CuO offline, 900 C
>lab. 3: +CuO offline, 750 C, 4 hrs
>lab. 4: +CuO offline, 600 C, 6 hrs
>lab. 5: element. anal.
>
>(guess which one's my lab....)
>
>The results are corrected for differences in other (approved)
>standards where those data were supplied by the participants.
>
>SUMMARY OF RESULTS
>
>lab no.      no. spls.     mean (ppm)      95 % conf. interval (+-)
>
>1               5               -24.53                0.06
>2               3               -25.15                0.05
>3               2               -24.11                0.38
>4               7               -24.45                0.05
>5               7               -24.83                0.15
>
>All values taken together produce a mean of -24.64 ppm +- 0.13 ppm..
>
>This is much better than numbers published by Rozanski et al. (1992,
>Radiocarbon 34(3), 506-519). For C3 a value of -24.91 +- 0.49 (!) was
>given there. The precision of intercomparison tests for other
>standards (C1 to C6) was also in that range.
>
>As one of the participants pointed out, he was "astonished at the
>wide error bands of carbon" in that publication. He remarked that
>"there is clearly a need to check the carbon scale as it is already done
>for hydrogen and oxygen." I can only agree.
>
>The results of this little test have been assuring, not only for the
>overall homogeneity of C3 in terms of stable carbon isotopes, but
>also for my use of the low combustion temperature. It is, by the way,
>interesting how many slightly different temperatures are used for
>cellulose combustion. It seems not to make any difference, anyway...
>
>Frank
>
>---------------------------
>Frank Pawellek
>Stable Isotope Laboratory
>Department of Geography
>University of Wales, Swansea
>Singleton Park
>Swansea SA2 8PP
>Wales / GB
>
>Tel. +44 1792 295148
>Fax. +44 1792 205556
>email [log in to unmask]
>------------------------(;->