Apparently, the rest of you think the -34 reading in Bergalnd on 4/1/1923 is legit. It is just a bit questionable that even the December record low is only -30 set in 1983. Also, Bergland typically does not seem to get as cold as Humboldt in a given cold wave.
For the sake of accuracy - you're looking at the modern "Bergland Dam" station that only extends back to 1948. This station doesn't appear to be as much of a cold spot as the original Bergland. The original Bergland also recorded -29 on 3/27/1913, so readings in the -30 range are (were) definitely possible at that time of year.
The original Bergland station existed from 1909-1925.
Also, regarding the comparison to December - the month of December is not a big cold wave producer in the upper Midwest. Their winter season really peaks after January 1st. Between 1909-1925, the lowest December reading in the entire state of Michigan was -38. Meanwhile, March had dropped to -40 or lower in three separate years (1913, 1917, 1920).
Having said all that, it's certainly possible that the -34 at Bergland on 4/1/1923 was erroneous. The thermometer may have been placed too low to the ground, or alcohol condensation in the bulb caused an erroneously low reading (a known problem at the time), or the observer was simply fibbing in order to make his numbers look more impressive. All three of those things used to happen at COOP stations.
However, given that Humboldt was -30 on 4/1/1923, Mio on the lower peninsula was -23, and three different stations were between -30F and -32F across Lake Superior in southern Ontario that morning, I don't think there's enough evidence to outright dismiss the Bergland reading. Just my opinion.