The United States and Baker Aircraft Sales, Inc. (Baker), appeal from a judgment rendered against them; Mrs. Furumizo cross-appeals, claiming that the damages awarded are inadequate. The decision of the trial court is reported at D.C., 245 F.Supp. 981. It sets out the facts, most of which are not disputed, at length. In this opinion, we do not restate them, but refer to the findings of the trial court by the numbered paragraphs of its decision.

The United States was held liable under the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671 et seq. Baker was sued and held liable under diversity jurisdiction, 28 U.S.C. § 1332. The action arises from the death of appellee Betty K. Furumizo’s husband in an airplane accident that occurred at Honolulu International Airport. We consider the appeals separately.

I. The Appeal of the United, States

a. The claim that the negligence of Baker’s flying instructor was the sole proximate cause of the accident.

This argument is in the alternative, the United States claiming that the negligence of Shima, the flying instructor, either superseded that of the United States’ air controllers or was the sole proximate cause of the accident. We do not find merit in either argument. In considering them, we assume that the controllers were negligent, as the court found, and that Baker was negligent, either in furnishing an inadequately trained instructor (Shima) or by reason of Shima’s own negligence. As is shown by findings 30, 31 and 32, the accident occurred within less than a minute after the Piper plane was cleared for takeoff. To say that, under these circumstances, the negligence of Baker in providing an inadequately trained instructor, or that of Shima in taking off, superseded the negligence of the controllers in encouraging or permitting him to take off, seems to us to be quite unrealistic. And it is equally unrealistic to say that the controllers’ negligence was not a proximate cause of the accident. In this respect, we agree with the trial court’s findings 38, 42 and 55. We certainly are not convinced that they are clearly erroneous. See Mitchell v. Branch & Hardy, 1961, 45 Haw. 128, 132, 138-139, 363 P.2d 969, 973, 976-977; United States v. Miller, 9 Cir., 1962, 303 F.2d 703, 711.

b. The claim that the United States was not negligent.

The trial court held that the United States was negligent on two theories. One is embodied in paragraphs 64-68 of its decision. In essence, it is that the tower controllers had a duty to go beyond the prescriptions of the Air Traffic Regulations and the Air Traffic Control Procedure Manual, and to withhold or delay clearance to a small plane, such as the[*968] Piper, when its position is such that it might take off into the wake turbulence of a large plane; that the mere warning prescribed in the manual is not enough. This theory we neither accept nor reject, because the second theory is, on the facts of this case, sufficient to sustain a finding of negligence.

The second theory is that, in this particular case, the controllers, having given a warning, actually saw the Piper start to take off in apparent disregard of that warning without waiting long enough for the wake turbulence to dissipate, and did nothing to stop it. This theory appears in paragraphs 39, 40, 41 and 42 of the trial court’s decision. It is fully supported by the evidence. The government gives little attention to this second theory in its brief. Its view is that the controllers did all that the regulations and the manual required them to do when they issued the clearance reading “caution turbulence departing D C eight cleared for take-off.” The government maintains that giving this type of clearance transferred to the pilot of the Piper, or left with him, the sole responsibility for avoiding the danger. No doubt, he had that responsibility. United States v. Miller, supra. And it may be (we do not decide) that if, after giving the warning, the attention of the controllers had been diverted elsewhere, either by their duties or even fortuitously, so that they did not see the Piper start its takeoff in disregard of the warning, the United States would not be liable. But we are unwilling to hold, as the government would have us do, that when the controllers did see the Piper start its takeoff, they had no duty to act. The danger was extreme, and they knew it. Nothing in the regulations or manual says that, under such circumstances, controllers shall not act. As the trial judge points out (paragraphs 61, 62, 63) the regulations and manual do not make mere automata of the controllers. Their job requires that they act in the interest of safety, and it would be strange indeed if that overriding duty did not include an obligation to seek, by appropriate instructions, to warn a pilot who is starting to take off when it is apparent to them that he will encounter a severe hazard. The government argues that because the manual provides that “when controllers foresee the possibility that departing aircraft might encounter * * * thrust stream turbulence or wing tip vortices from preceding aircraft” they should issue “cautionary information to this effect * * * to the pilots concerned,” the controllers are not required to do- more. But we do not think this directive is fully complied with where, although a first warning has been given, it becomes clear to the controller that another warning is needed and none is given.

Applicable here is the rule announced in Restatement of Torts (Second) § 321:

“Duty to Act When Prior Conduct is Found to be Dangerous.

(1) If the actor does an act, and subsequently realizes or should realize that it has created an unreasonable risk of causing physical harm to another, he is under a duty to exercise reasonable care to prevent the risk from taking effect.

(2) The rule stated in Subsection (1) applies even though at the time of the act the actor has no reason to believe that it will involve such a risk.

Comment:

a. The rule stated in Subsection (1) applies whenever the actor realizes or should realize that his act has created a condition which involves an unreasonable risk of harm to another, or is leading to consequences which involve such a risk. The rule applies whether the original act is tortious or innocent. If the act is negligent, the actor’s responsibility continues in the form of a duty to exercise reasonable care to avert the consequences which he recognizes or should recognize as likely to follow. But even where he has had no reason to believe, at the time of the act, that it would involve any unreasonable risk of physical harm to another, he is under a duty to exercise reasonable care when, because of a[*969] change of circumstances, or further knowledge of the situation which he has acquired, he realizes or should realize that he has created such a risk.”

We note that Hawaii has followed the Restatement in Mitchell v. Branch & Hardy, supra, 45 Haw. at 132, 137, 363 P.2d at 973, 976. We need not discuss the cases on which the government relies, as they are expounded at length in the trial court’s opinion. None, we think, is contrary to the decision here, as far as the trial court’s second theory of negligence is concerned.

II. The Appeal of Baker

a. The claim that Baker was not negligent.

The trial court found Baker liable on two theories. The first is that Baker negligently provided an inadequately trained instructor. The findings based upon this theory appear in paragraphs 36 and 38. The second theory is that Shima was himself negligent in taking off when he did. The second is the only theory on which the case was tried, argued and submitted. Baker correctly argues that the first theory was not presented in the pleadings or the pretrial order, and was not argued. It first appeared in the trial court’s decision of June 21, 1965. On June 30, 1965, Mrs. Furumizo moved to amend the complaint and pretrial order to state the first theory and this motion was granted. Baker now vigorously argues that it should not be held liable on the first theory because it had no notice of it and therefore had no opportunity to meet it. It argues that it could have produced additional evidence that might have convinced the trial judge that Shima was fully and properly instructed as to the dangers of taking off in a small plane into the wake turbulence of a very large one. We are inclined to agree, although we do not decide this question.

Baker also argues that the court made no finding on the second theory, the one on which the ease was tried. This is literally correct. Nowhere is there a direct finding that Shima was negligent. The trial judge thought that he had made such a finding, as is indicated by a colloquy between court and counsel on motion for a new trial. In that the court was mistaken, and such a situation would normally require a remand for further findings on the issue. Here, however, it is apparent from the record that the court intended to make the finding; presumably it would do so on remand. We think that such a remand is unnecessary here. All that Baker seeks, in relation to the first theory, is an opportunity to convince the trial judge that Shima was in fact fully informed about the danger of wake turbulence, thus showing that Baker was not negligent in employing him as an instructor. But if Baker succeeded in proving that, it would bring itself within the second theory, the one on which the ease was tried. As the trial court said (paragraph 38):

“Had Shima been an adequately trained pilot, and had he been adequately instructed and made fully aware of the dangers of large plane wake turbulence to small planes, the Court cannot believe that he would have taken off at the time and under the circumstances he did, the evidence indicating that he was a careful and prudent person, not one given to taking undue chances. A well-trained pilot, adequately schooled in the actual dangers to planes of the size of this Piper of turbulence created by planes of the size of the DC-8, would have done what the pilots of the C-47 did — wait until a reasonable time had elapsed to allow the most violent aspects of the turbulence to dissipate before taking off * * * ”

To have done anything else would certainly have been negligent. In substance, then, the trial court’s findings do embody the second theory. Remand for a more specific finding would be a triumph of form over substance.

b. The claim that decedent Furumizo was guilty of contributory negligence.

The trial court found that he was not, the finding on this issue being con[*970] tained in paragraph 37. This finding is based on the proposition that Shima, as instructor, was in control of the takeoff, which he clearly was. It is also based on the failure of the defendants to show that Furumizo knew the danger. On this question, the evidence is in conflict, and we cannot say that this finding, embodied in the trial court’s footnote 5, is clearly erroneous.

c. The claim that the damages awarded are excessive.

1. The estimate of future earnings.

The court’s findings appear in paragraph 70. Baker first attacks sub-paragraph (c), in which the court determined that it was reasonably certain that Furumizo would have been promoted to Grade GS-11 within 10 years. It is urged that this is pure speculation. In a sense, of course, any prediction of this kind is speculative. If things were not as they are, they would be quite otherwise, but no one can be certain as to just how they would be otherwise. Nevertheless, in a wrongful death case, loss of future earnings must be estimated. In that sense, the court must speculate. All that is required is “reasonable certainty,” Condron v. Harl, 1962, 46 Haw. 66, 72, 374 P.2d 613, 617, an uncertain phrase used to limit what must inevitably be a decision lacking in certainty. It is not the law, as Baker seems to suggest, that the court cannot take into account probable promotions. Ginoza v. Takai Electric Co., 1955, 40 Haw. 691, 707, and see Rohlfing v. Moses Akiona, Ltd., 1961, 45 Haw. 373, 369 P.2d 96. The finding is not clearly erroneous.

Next, Baker attacks subparagraph (h) on the ground that, in multiplying present value of earnings between June 20, 1961, and June 20, 1965, by a factor of 1.15865 so as to eliminate the discount to present value as to that part of the earnings, the court did not apply the same factor to the deductions of income taxes (subparagraph (i)) and pension contributions (subparagraph (k)), for that period. Logically, Baker is correct. However, we must remember that the ultimate total damage figure awarded is the sum of a series of predictions, none of which involves mathematical certainty, and that it is the reasonableness of the ultimate figure that is really in issue in such a case as this. Since we are of the opinion that the final figure arrived at is not unreasonable, we decline to require the trial judge to modify it as Baker urges. Were we to do so, we think that we would be required to reopen the whole question of damages.

Third, Baker urges that the trial judge calculated decedent’s personal expenses as 35% of salary less pension contributions and income taxes, rather than 35% of gross, whereas Mrs. Furumizo’s estimates applied the 35% to the gross. The trial judge was not bound by this approach, and we cannot say that his approach is unreasonable. See subparagraph (k).

Finally, Baker says that $15,000 damages for decedent’s pain and suffering is too much, because death was almost instantaneous (paragraph 71(a)). We do not find the award shocking. The question was primarily for the trial judge and we cannot hold that he erred. See Rohlfing v. Moses Akiona, Ltd., supra.

III. The Cross-Appeal of Mrs. Furumizo-.

Her cross-appeal is limited to the claim that the damages awarded are inadequate. What we have said about Baker’s attack upon the damages awarded is equally applicable here.

First, Mrs. Furumizo says that it was error for the court to decline to take into account a probable future decline in the value of the dollar, in line with what has happened to it in recent years. We think that the record does not support the contention. It is true that the court made no computation of probable decline. But in making its other computations, it gave Mrs. Furumizo the benefit of the doubt in computing probable income taxes, in assuming the continuance of the cost of living allowance for federal employees in Hawaii, in omit[*971] ting computation of state income taxes, and in allocating to her a high portion of the award (paragraph 70, subparagraphs (d), (e), (i), (j)). Moreover, the period here in question is 34.6 years (paragraph 70(a)) and this court knows that less than that long ago there was a period of severe deflation. We agree with the court’s finding 70(g).

Second, it is claimed that the court erred in not allowing prejudgment interest. But the court allowed this, in effect, by not discounting the award for the period from death to judgment. See Sisson v. Weathermon, 1961, 252 Iowa 786, 108 N.W.2d 585. Moreover, no Hawaii case has been cited that allows such interest. The Hawaii statute speaks only of interest on the judgment. R.L.H. 1955, § 191-2. In any event, the United States is not liable for such interest. 28 U.S.C. § 2674.

Third, it is argued that Mrs. Furumizo’s loss should not be computed as a portion of after tax earnings. Reliance is upon Kawamoto v. Yasutake, 1966, 49 Haw. 42, 410 P.2d 976. But it is not in point. It dealt with an attempt to have a jury told, in fixing damages for personal injury, that the award is not taxable. Here, the question is, what support would have been available to Mrs. Furumizo from her husband’s earnings, if he had lived. The old saw about the certainty of taxes is still good. It is reasonably certain that what would have been available would have been after-tax dollars, not pre-tax dollars. See Southern Pacific Co. v. Guthrie, 9 Cir., 1951, 186 F.2d 926. We think that, at the least, the court could take this fact into consideration, although it might not have erred if it had refused to do so. See United States v. Becker, 9 Cir., 1967, 378 F.2d 319, decided May 16, 1967, involving Arizona law. We come back to the basic proposition that the total award is but an estimate based on a series of predictions, and that the ultimate question as stated in the Hawaii statute, R.L.H.1955, § 246-2 is, what is “fair and just compensation, with reference to the pecuniary injury * *

Fourth, it is said that the court should not have deducted, from the award to the decedent’s estate, the $250 burial cost paid by the Veterans Administration (paragraph 71(a)). The short answer is that Mrs. Furumizo, in her capacity as administratrix, has not appealed.

Fifth, it is said that the judge, in awarding damages for loss of consortium, etc. (paragraph 71(d)), abdicated his function because he thought it proper to “try to put himself in the place of average citizens and jurymen, and try to reach an amount that he has reason to believe would approximate what such a jury in his community, without undue emotional influence, would award as reasonable.” The argument is without merit. We daresay that many a judge makes the same effort, the only difference here being that in this case the judge said so. The fact remains that the decision is that of the judge, and of no one else. We are not here to flyspeck his mental processes.

Finally, it is argued that the general damages are too low, because the judge took into account Mrs. Furumizo’s health, attractiveness, youth, prospect of remarriage, and earning capacity (paragraph 71(d)). All of these, and many more, are things that either a judge or a jury might think about, and that the judge did think about, in awarding general damages for “loss of love and affection, including (a) loss of society, companionship, comfort, consortium or protection, (b) loss of marital care, attention, advice or counsel” (R.L.H.1955, § 246-2). What we have just said about the fifth contention is equally applicable here.

Affirmed.