Recent research on international productivity comparisons with historical data has encountered large discrepancies between benchmark comparisons and time series extrapolations from other benchmarks. Broadberry and Burhop (2005) have recently argued that for Hoffmann's (1965) widely accepted time series for German industrial output, there is no such productivity paradox, while for a revision of that series recently suggested by Ritschl (2004), the discrepancy between the Anglo-German benchmark and the time series projection is considerable. Attempting to reconcile the time series evidence and the productivity benchmarks, they discard the revised series in favor of the original, disregarding mounting evidence on its lacking reliability. The present paper restates this productivity paradox and proposes a possible resolution. We draw on recent archival discoveries by Fremdling and Staeglin (2003) and Fremdling (2005) that confirm the revisions to the Hoffmann series. We also draw on recent advances in the reconstruction of a German industry census of 1936, and argue that the productivity paradox is largely the consequence of mismeasurement in all versions of the German series. Correcting for the omissions, much of the Anglo-German productivity paradox disappears.