This is an action brought under the Federal Employers’ Liability Act, 45 U.S.C. §§ 51-60, to recover damages for injuries sustained by plaintiff McWeeney on March 23, 1956, while he was employed by defendant New Haven as a yard brakeman. McWeeney was struck by a moving freight car while he was on the side ladder of another ear. There was conflicting evidence whether he was contributorily negligent and, if so, in what degree, and also whether he was totally or only partially disabled. He was a bachelor, aged 36 at the date of the accident and 39 at the time of the trial. His pay for the 13 weeks immediately preceding the accident was $1,187.61, an annual rate of approximately $4,800. The case was tried to a jury which awarded him a verdict of $87,000. Defendant’s motion for a new trial was denied. Defendant appealed, claiming the trial court erred in denying ten requests to charge.
The appeal was argued to a court consisting of Chief Judge LUMBARD, Judge MOORE and the writer. We were unanimous in finding no error in the refusal to give the five charges requested with respect to defendant’s negligence and the three sought in regard to plaintiff’s contributory negligence. However, as will appear from the opinions herein, we were in disagreement as to the judge’s refusal to grant Requests No. 17 and No. 18, namely:
“17. If you arrive at a verdict under the Court’s charge in favor of plaintiff, you will not add any sum of money to the amount of the verdict on account of federal or state income taxes, since the amount awarded to the plaintiff by your verdiet is not taxable income to the plaintiff within the meaning of these tax laws.”
“18. If your verdict is in favor of plaintiff, you must calculate any past or future loss of earnings on the basis of his net income after deduction of income taxes.”
Since similar issues arise frequently in trials in this Circuit, we referred the case to the court in banc, 28 U.S.C. § 46(a), which has considered it on the record and briefs previously submitted. We shall discuss the two requests in inverse order.
I.
The proposal that juries in personal injury or death cases should always be instructed to excise from any recovery for loss of earning power the amount that plaintiff or his decedent would have been required to contribute to the fisc, has a surface appeal,1 although it has not won assent from most courts that have considered it.2 We join the majority, with a qualification set forth below.
It is not altogether clear whether the proponents of the instruction say that the tax adjustment is simple for a jury, that, although it is difficult, the jury’s assessment of loss of earning power is already so complicated that Ossa may as well be piled on Pelion, or just that there is no burking the problem whatever the difficulties may be. We think none of these contentions is sound in the usual case.
Only three elements need now be found in determining damages from loss of earning power in a total disability case — future normal earning power, expectancy, and discount factor.3 The third is charged by the court, often accompanied, as it was here, by a rule of thumb formula to simplify the jury’s task. Expectancy is usually determined *36by life tables and again is normally charged by the court, although a more refined analysis would permit reduction of the figures in the tables to take account of “the probable duration of plaintiff’s earning capacity” (presumably with due regard to pension rights thereafter), 2 Harper & James, The Law of Torts, § 25.8, at 1317 and fn. 6 (1956), and other authorities cited in Conte v. Flota Mercante del Estado, 2 Cir., 1960, 277 F.2d 664, 670, and also of conditions unconnected with the accident that might have reduced plaintiff’s expectancy below the normal term. American Law Institute, Restatement of Torts, § 924, Comment e. This leaves future earning power. As to that, of course, there are great imponderables, since existing compensation is only one element in the scale. But, whatever difficulties may inhere in this element of the formula, they do not justify the addition of other much more serious ones in the ordinary case.4
The instruction here requested and refused illustrates the delusive simplicity with which the subject has been invested. Defendant wished the jury to be told merely that “If your verdict is in favor of plaintiff, you must calculate any past or future loss of earnings on the basis of his net income after deduction of income taxes.” All of us agree the jury needs more guidance than that. But what is the alternative? The trial court could begin by instructing that under the optional tax provided in § 3 of the Internal Revenue Code, which McWeeney was entitled to elect, the tax on a bachelor’s income of $4,800 would have been $773. This is still simple enough but the jury must determine not what Mc-Weeney’s tax on $4,800 now is but what it would be over his expectancy. In these lower brackets the amount of the tax and its percentage relation to earnings are enormously affected by the number of exemptions. The simple act of matrimony, coupled with the filing of a joint return, would reduce McWeeney’s tax from $773 to $620. Is the jury to consider the likelihood of this not unusual occurrence? If the lady brought two children with her, or if these were produced in the ordinary way, the tax would be cut in half, to $380. Each additional child would bring a further tax saving of $110, so that a total of five would make the tax nominal. While such fecundity might be unlikely in a plaintiff of 39 the rule here framed for McWeeney must apply to men who evidence greater interest in marriage and parenthood, and the rise in the birth rate is a phenomenon of our age.5 Is the jury in each case to speculate, or hear testimony, on the procreative proclivities and potentialities of the plaintiff and his spouse? Moreover, children are by no means the only source of exemptions; § 152 of the Code lists nine other categories. Nor will it do to say that this is over-refinement. If a defendant claims that a plaintiff in these brackets would have had to pay 20% of his income in taxes, the court cannot deprive the plaintiff of an opportunity of showing that he would not have paid anything of the kind, and once the rule were adopted, we see no way of eliminating the question of potential babies and the exemptions they trail with them.6
*37Having duly instructed the jury how to estimate the number of plaintiff’s probable exemptions, and the dates when these will come into being, and on rate brackets and deductions, and the treatment of other income of the plaintiff not affected by the accident,7 the court would necessarily encounter another problem. Even the best designed instruction that stopped at that point would be unfair to the plaintiff. This is because, as proponents of the instruction recognize, the product of plaintiff’s lost earning power ex-tax and his expectancy will have been discounted to produce “that sum of money which if invested at a fair rate of return will yield annually the amount by which the plaintiff’s earning capacity has been lessened and which will at time of end of the plaintiff’s life expectancy be reduced to zero. This takes into account the fact that money earns interest each year; and it should be remembered that this interest is taxable. Therefore, if a court is going to use income after taxes as a measure of plaintiff’s loss, it must add back the taxes which would be due on the interest earned — else the award would not fully compensate for the loss.” 8
All this is simple enough to state but how is the jury to apply it? If we suppose that plaintiff would purchase an annuity with the lump-sum previously determined, it will hardly help the jury to be told, in the language of § 72(b) of the Internal Revenue Code, that “Gross income does not include that part of any amount received as an annuity under an annuity, endowment, or life insurance contract which bears the same ratio to such amount as the investment in the contract (as of the annuity starting date) bears to the expected return under the contract (as of such date).” Of course the court could itself calculate the exclusion ratio and charge what proportion of the annual payments would be taxable, thus leaving to the jury only the task of determining the amount of each annual payment that would be taxable, the tax thereon, and the discounted amount of the sum of such tax payments, and adding this to the ex-tax award. But here a new complication seems to arise. For a portion of the periodic payments on this sum, which is to be added to fund plaintiff’s tax liability, would itself be subject to tax.
It is answered that, however unrealistic it may be to suppose that a jury could properly make the series of calculations required,9 the worst result from giving an instruction would be better than the best result from not giving one. This ignores that, by imposing on the jury a task that the jury cannot reasonably be expected to perform, we would be likely to impair the quality of its performance in areas of its true competence. It ignores also that although failure to instruct the jury to reduce a verdict to take account of income taxes on plaintiff’s lost earning power, when viewed in isolation, may tend to make verdicts too high, there are at least two *38other factors that countervail this in the ordinary case.10
The first is inflation. Though some courts have sanctioned instructions permitting the jury to take into account inflation between the injury and the trial, there is little or no authority in favor of charging the jury to take future inflation into account, see 2 Harper and James, The Law of Torts, § 25.11 (1956). Yet there are few who do not regard some degree of continuing inflation as here to stay and would be willing to translate their own earning power into a fixed annuity,11 and it is scarcely to be expected that the average personal injury plaintiff will have the acumen to find investments that are proof against both inflation and depression — a task formidable for the most expert investor. The effect of inflation of 1% a year over McWeeney’s 29-year expectancy at trial would go a long way toward offsetting any excess in the verdict due to failure to deduct income tax.12
The second offsetting factor is that the supposedly overcompensated plaintiff does not retain his entire recovery or anything like it. Whatever the reasons of history or policy for the American practice of generally not awarding attorneys’ fees to the successful party, see Goodhart, Costs, 38 Yale L.J. 849, 872-78 (1929), we can hardly shut our eyes to this when asked to require the jury to take another extrinsic factor into account — particularly when we know that even court-prescribed maximum scales of contingent fees, which have been attacked by counsel as inadequate, provide either a sliding scale ranging from 50% down to 25% or a flat amount of 33% %, N.Y.App.Div. 1st Dept., Special Rules Regulating the Conduct of Attorneys and Counsellors at Law, Rule 4.
There may be cases where failure to make some adjustment for the portion of a plaintiff’s or decedent’s earnings that would have been taken by income taxes would produce an improper result; but these are at the opposite end of the income spectrum from McWeeney’s. For example, if a plaintiff or a plaintiff's decedent, had potential earnings of $100,000 a year, more than half of which would have been consumed by income taxes, an award of damages based on gross earnings would be plainly excessive even after taking full account of the countervailing factors we have mentioned.13 We find it hard to believe that juries would render such a verdict even in the absence of instruction; but in this limited class of cases the court may properly give some charge or, perhaps better, use the tools provided by Fed.R.Civ.Proc. 49, and an excessive verdict may always be set aside. In such cases, which in proportion are relatively few, the criticism that the whole process of computa*39tion is unrealistic has a considerable measure of validity, the projection of future income at such levels being itself extremely conjectural, and the slope of the tax progression curve declines although only after having reached such a high plateau that earnings above it have relatively slight value. Such cases are in sharp contrast to the great mass of litigation at the lower or middle reach of the income scale, where future income is fairly predictable, added exemptions or deductions drastically affect the tax and, for the reasons indicated, the plaintiff is almost certain to be undercompensated for loss of earning power in any event. To be sure a practice of refusing the instruction in most cases but requiring some adjustment in a few lacks precision and elegance; but we cannot disregard the practicalities of judicial administration and “the law is untidy as life with which it deals.”14 Just where the line should be drawn must be left, as so much is, to the good sense of trial judges, whether acting with a jury or without. Certainly the line was not approached here.
We do not consider that O’Connor v. United States, 2 Cir., 1959, 269 F.2d 578, 584-585, commits us to the course sought by defendant. O’Connor was an action under the Federal Tort Claims Act by a v/idow to recover for the loss sustained by herself and a minor child from the death of her husband. Plaintiff’s rights were governed by Oklahoma law. The government argued that “Oklahoma has thus joined the jurisdictions which hold that, if damages are supposed to be of a purely compensatory nature, they must be based upon the net earnings of the decedent, his take home pay,” citing Magnolia Petroleum Co. v. Sutton, 1953, 208 Okl. 488, 497, 257 P.2d 307, 316; and the Court agreed. Although our much cited decision in Stokes v. United State's, 2 Cir., 1944, 144 F.2d 82, 87, denying a deduction for income taxes in an action against the United States under the Suits in Admiralty Act, 46 U.S.C. § 742, as “too conjectural,” was discussed in both briefs in O’Connor, the Court did not overrule it. We continue to adhere to Stokes where the question is one of federal law or the applicable state law is silent, subject only to the qualification stated in the preceding paragraph.
II.
Request No. 17, which Judge Weinfeld also declined to give, was to instruct the jury not to “add any sum of money to the amount of the verdict on account of federal or state income taxes, since the amount awarded to the plaintiff by your verdict is not taxable income to the plaintiff within the meaning of these tax laws.” This correctly states the law, Int.Rev.Code of 1954, § 104; N.Y.Tax Law, § 359(2) (e). Unlike Request No. 18, it imposes no new burdens on the jury and there is nothing speculative about it. Hence there would have been no error in the court’s giving the instruction. The question before us is not that but whether the failure to give it was error, and error so serious as to require a new trial.
We do not think so. The requested instruction was that the jury should not add something to the recovery' because of a factor for which the plaintiff had never suggested it should. Be-fore an appellate court should hold that failure to give such a cautionary instruction was reversible error, there ought to be evidence either that juries in general increase recoveries on this account or that the particular jury did so. The published material on the former point is too slender to support a judgment that juries do this generally,15 and there is nothing' to suggest that this one did. McWeeney had incurred medical expenses of $10,305 (only $2,065 of which was challenged *40in any way), had sustained and would sustain pain and suffering, had lost some $14,000 in earnings up to the date of the trial, and had a future earning potential which, discounted on the basis that the judge charged without objection, was nearly $100,000. Under these circumstances, a verdict of $87,000 affords no indication that the jury included an allowance for income tax that McWeeney would never have to pay. To be sure, one may speculate that the jury may have found McWeeney guilty of contributory negligence or may not have believed his testimony as to total disability, and therefore first reduced the amount and then increased it for the non-existent income tax liability; but this is sheer conjecture and no such claim was advanced in support of defendant’s motion for a new trial. The accident occurred over four years ago; the case was tried before an experienced judge of the most scrupulous fairness; and we see no sufficient reason for ordering a new trial because he declined to give a cautionary instruction when there is nothing to indicate that one was needed.
Judgment affirmed.
. Harper & James, Tie Law of Torts, § 25.12 at 1326-27.
. The cases are cited in Morris and Nordstrom, Personal Injury Recoveries and the Federal Income Tax Law, 46 A.B.A. Journal 274, 276, fn. 19 (1960).
. In a partial disability ease there is a fourth, plaintiff’s post-accident earning power.
. We know all too little about how juries actually deal with this and other problems in fixing damages in personal injury cases. Happily, this deficiency seems in a fair way to be remedied by the forthcoming publication of the studies made by the Jury Project of the Law School of the University of Chicago. See Kalven, The Jury, the Law and the Personal Injury Damage Award, 19 Ohio St.L.J. 158,159 (1958).
. See the figures in W. W. Rostow, The Stages of Economic Growth (1960), pp. 80-81.
. See Southern Pacific Co. v. Guthrie, 9 Cir., 1951, 186 F.2d 926, 927, certiorari denied, 341 U.S. 904, 71 S.Ct. 614, 95 L. Bd. 1343, where the court said on a rehearing in banc, “We think the [district] court’s view that the net take home pay, after taxes, would represent the actual loss, is correct; but wo are now convinced that we cannot tell how much this will be.” This was because the plaintiff “could look forward to an additional exemption after age 65, and because he was married, the split income features of the law would give two additional exemptions when his wife reach*37ed 65, something about which we cannot tell.” If an appellate court is thus unable to determine “the net take home pay” of a plaintiff aged 58 (where, in fact, it would seem that actuarial tables could have been so used as to provide an answer), what is a jury to do with one thirty years younger ?
. It has been proposed that this be ignored, see Morris and Nordstrom, supra note 2, p. 328.
. Morris and Nordstrom, supra note 2, at 328. See also Nordstrom, Income Taxes and Personal Injury Awards, 19 Ohio St.L.J. 212, 227-28 (1958). '
. See Armentrout v. Virginia Ry., D.C.S. D.W.Va,1947, 72 F.Supp. 997, reversed on other grounds, 4 Cir., 1948, 166 F.2d 400, 4 A.L.R.2d 1064, where the court’s eight pages of calculations give an example of what the jury might be asked to do.
We have not dealt with the added problems relating to state income taxes, which the proponents of the instruction, quite logically, seek to have included. Mc-Weeney had first worked in Connecticut, then in Florida. Was the jury to determine whether McWeeney would have remained a New Yorker or moved to a state without an income tax — and when?
. We have omitted mention of a third possible factor, namely, that juries do not usually award the full amount that the earning power X expectancy — discount formula would demand. See Kalven, supra, note 4, at 161-62.
. Hence the demand for variable annuities. See S. B. C. v. Variable Annuity Co., 1959, 359 U.S. 65, 70, 100, 79 S.Ct. 618, 3 L.Ed.2d 640.
. In fact, the annual, rate of depreciation of the dollar (compounded) from 1949 to 1959, measured by reciprocals of official cost-of-living or consumer price indexes, was 2.0%. Even for 1954-59 it has been 1.7%. The First National City Bank of New York, Monthly Letter, July, 1960.
. The decision of the House of Lords in British Transport Commission v. Gour- • ley, (1956) A.C. 185, was such a case. The plaintiff was 69 years old at the date of the trial. The award for loss of earnings was £37,720, which the tax deduction reduced to £6,695. The case had been tried to a judge. Although the House of Lords contemplated application of the principle also in jury trials, as a practical matter the English courts have accomplished “the virtual abolition of the jury system in all civil actions which are not concerned with the personal reputation of the litigants.” Good-hart, Current Judicial Reform in England, 27 N.Y.U.L.Rev. 395, 407 (1952). And, of course, the successful plaintiff in England recovers from the loser all reasonable costs.
. Felix Frankfurter Reminisces (1960), p. 21.
. Kalven, A Report on the Jury Project, Conference on the Aims and Methods of Legal Research, University of Michigan Law School, 167-68 (1957).