Frank Lasner dug this up. DMI has old ice extent maps on their web site, like this one from August 1938.
Hide the decline – Latest News (hidethedecline)
brunnur.vedur.is/pub/trausti/Iskort/Pdf/1938/1938_08.pdf
It shows the Northeast Passage nearly open, with several hundred miles of open water north of Russia. Now compare vs. August 1996 – when the Northeast Passage was frozen up.
The image below overlays the 1996 map on the 1938 map. Green represents more ice in 1996, red shows more ice in 1938. There was more ice in August 1996 than in August 1938.
So here is the problem. UIUC has this graph on their website, which shows a lot more ice in summer 1938 than in summer 1996.
seasonal.extent.1900-2010.png (1296×1057)
That isn’t the first problem found with this graph. It also shows 1974 extent higher than 1979, which we know from the 1990 IPCC report to be incorrect.
www.ipcc.ch/ipccreports/far/wg_I/ipcc_far_wg_I_full_report.pdf
Bottom line for me is that I don’t trust any of the pre-satellite ice or temperature data being published now. They appear to be political documents rather than historical.