Original Structure:   Magn. Reson. Chem., 43, 264 (2005)
Revised Structure:    Magn. Reson. Chem., 44, 571 (2006)

This example has been brought to my attention by Heinz Kolshorn from University of Mainz - both datasets (and also both articles) under discussion here, are not within my internal CSEARCH-environment as per March 28th, 2008!

The following C-NMR data (172.9/S, 70.1/D, 69.6/D, 62.7/T, 43.2/Q and 38.3/T) have been used to elucidate compound #1 (numbering scheme from MRC-2006) as 3,6-dihydroxy-N-methyl-2-piperidone. The investment of approximately 10msec of CPU-time reveals an average deviation of 12.5ppm between experimental CNMR-data and the predicted shift values as shown below. 3 Carbons (out of 6!) have deviations between 11 and 34ppm !

Original structure

The revised structure by T. Winkler leads to a known proline-derivative; the average deviation between experimental and predicted carbon chemical shift data is now 1.8ppm as shown below.

Revised structure

All predictions have been done using CSEARCH-NN-technology ommitting solvent-dependant prediction. Compound #1 has been predicted without stereochemical information, whereas the prediction of compound #3 was done using stereochemical information. Both structures (and also both articles) are not avaialable within all the data at my internal CSEARCH-installation !

This example leads directly to the central problem of our publishing procedures:

The facts are:


Apply spectrum prediction whenever possible

Page written by:   Wolfgang.Robien(-at-)c13nmr.at
Page written on:   March 28th, 2008