As previously mentioned, the coldest airmass so late in the season is moving into Fairbanks-land. This is based on the period of record for upper air soundings, which is this case is since 1948. The WB/NWS in Fairbanks has taken has upper air data since the late 1930s but I've been unable to locate the actual observations from before 1948.
There are numerous ways to assess the temperature of an airmass. In general it's best to use thickness, as this is proportional to the mean temperature through a given layer of the atmosphere. This time of year though it's simplest to use the 850mb temperature (roughly 4000' MSL), since the atmosphere is usually pretty well mixed from the ground up to that level.
For the second half of May and early June, here are the lowest 850 mb temperatures at Fairbanks after May 15th (since 1948). Notice the coldest is an estimate from the high resolution North American Regional Reanalysis as the RAOB was missed that day.
May 17 1992 00Z: -12.5C (RAOB missed, estimated from NARR)
May 16 1992 00Z: -11.7C
May 16 1992 12Z: -10.9C
May 19 1975 12Z: -10.9C
May 18 1964 12Z: -10.9C
May 18 1964 00Z: -10.5C
Jun 04 2006 12Z: -9.4C
Here is a plot of the GFS forecasted 850mb temperature from the Friday morning run:
Here's the same model forecast with the verification (through Sunday afternoon) plotted. Fairbanks (-14.3C) and Anchorage (-10.9C) both set records for lowest 850mb temps so late in the season.