(dissenting).
The opinion of Judge Kalodner on the petition for rehearing sufficiently removes from the case an apparent acceptance of the doctrine that a manufacturer has a continuing duty to develop improvements in the safety of his prod*243uct which was non-negligently produced. I would- therefore, join the concurring opinion of my brother Hastie but for the fact that the District Judge who sat without a jury expressly planted his decision on that ground and made no finding that the propeller system was defective at the time the defendant sold it or that defendant was negligent in failing to give warning of a defect. Thus the District Judge stated: “Assessing United’s development program for the correction of overspeeding problems against the background of the standard of care imposed upon it by law, I am forced to the conclusion that United was negligent in that it should and could have produced pitch lock for use on Constellation type planes prior to June 20, 1956 [the date of the accident], and had it done so, the ability of AMS substantially to control the overspeed would have been materially increased.” 219 F.Supp. at p. 572. Absent an express finding of fact of such negligence I find myself unable to concur in the majority’s conclusion.
It will not do to say, as does the majority on the petition for rehearing, that a finding of negligence on the date of the sale “is implicit in the trial court’s expressed finding that the defective design and manufacture caused the accident on June 20, 1956.” For the trial judge’s opinion itself recognizes that airplanes and their parts are still not perfect nor free from all defects. The question is not whether there was a defect in design and manufacture which caused the accident but whether such defect existed at the time of the sale, and was one which fell below the standard of due care under the circumstances and constituted negligence. Any other view would absorb negligence in the doctrine of absolute liability, which does not now require consideration as a preferable alternative.
Again, it seems to me impossible to hurdle the absence of a finding by the trial judge of negligence in manufacture or design at the time of the sale by declaring that we believe that such a finding could have been explicitly made from undisputed evidence.1 We may not make findings which were not made by the trial judge, even if the evidence was un-controverted.
It may seem an overemphasis on procedural niceties to insist that we may not rest our conclusion on a theory for which there are no supporting findings of fact. But the silence of the trial judge on this subject resulted from his adoption of a theory which is now excluded and we may not assume that on a new focus of attention the findings which may appear appropriate to us will inevitably be made by the trial judge. In any event, it seems to me that this is a matter in which we should not speculate. I therefore would vacate the judgment and direct the court below to make specific findings of fact on the charge of negligence at the time of sale and on alleged failure to warn of defects after the sale. The parties will have their opportunity to argue the inferences and conclusions to be drawn from the evidence, the trial judge who heard the witnesses will be able to make express findings on the facts, and we will not be required to become finders of fact in his stead to fill in the presently existing factual void.
I therefore dissent from the denial of the petition for rehearing.
. The majority opinion states: “In limiting our discussion to the issues as presented by the respondent we regretably did not declare that the undisputed evidence required an explicit finding by the trial judge that the respondent was negligent on July 15, 1955, when it supplied a propeller system of a design which it knew had earlier caused 10 serious ac-eidents of decoupling, fire or separation, and 57 accidents of lesser degree (engine fires or engine failures). The finding that the respondent was negligent on July 15, 1955, is implicit in the trial court’s expressed finding that the defective design and manufacture caused the accident on June 20, 1956.” [Footnotes omitted]