A review of the entire transcript, including not only the oral stipulations of counsel but also the evidence actually introduced and disputed, leads us to the conclusion that the only (and ultimate) issue which the parties cared to and actually did litigate was whether the plaintiff had "exercised due diligence with respect to maintaining heat in the building" within the meaning of paragraph 3 of the "Additional Exclusions" of the policy covering the building. As the case was tried, the only question of fact was whether there had been any fuel oil in the tank of the separate heating system for the vacant second-floor apartment of the building when the plumbing in that apartment froze and burst during the first week of January, 1972. The language and structure of the policy were such that the defendant had the burden of proving the negative of that issue. See Palmer v. Pawtucket Mut. Ins. Co. 352 Mass. 304 , 305, 306 (1967), and Ratner v. Canadian Universal Ins. Co. Ltd. 359 Mass. 375 , 381 (1971). On the evidence most favorable to the defendant the judge warrantably found that the tank was full on April 12, 1971, that the plaintiff had kept some heat in the apartment during the fall and winter of the 1971-1972 heating season until the time of the freeze-up, and that the next delivery of fuel had been made on January 28, 1972. Necessarily implicit in the judge's finding for the defendant on the ultimate issue was a subsidiary finding that the tank had been empty at the time of the freeze-up. That finding could not rest on the judge's disbelief of the plaintiff's testimony that the tank was one-quarter full on the day prior to the freeze-up. O'Connell v. Esso Standard Oil Co., Inc. 337 Mass. 639 , 642 (1958),
and cases cited. We are of the opinion that the subsidiary finding was clearly erroneous (see Marlow v. New Bedford, 369 Mass. 501 , 508 ) for lack of evidentiary support in the record. It appears to rest on nothing more than reasoning that a tank which was full on April 12, 1971, must have run dry by the first week of January, 1972. The difficulty with such reasoning is that there was no evidence of (a) the total storage capacity of the tank or (b) any of the other factors (degree days, normal efficiency of the burner, etc.) necessary to a determination whether the tank might have been emptied by the critical date. The defendant had, but did not sustain, the burden of proving that the tank was empty on that date. The judgment is reversed, and a new judgment is to be entered for the plaintiff in accordance with the stipulation as to the amount of the plaintiff's damages.