It's impossible. The water level can't fall below the cathode. F&P were very clear on this aspect (1): "The electrodes were held in place by the deep Kel-F plug and Kel-F spacer at the bottom of the cell." [underlines added]
You are mistaken. The cathode remained hot; the inside of the cell remained hot; the remaining water boiled away. If that had not happened, the Kel-F plug would not have melted. The remaining water was below the hot cathode, but in the enclosed space the cathode heated it enough to boil it away. You can easily confirm this by putting a resistance heater in a cell and . . . I joke! I don't mean it! Of course you will not confirm this!
Of course it did not boil away in calibration runs with electrolysis heat only.