OPINION
This wrongful death action is before us for the second time. On the first appeal, we announced our adoption of the doctrine of comparative negligence as the law of this state, and remanded so the superior court could apply that doctrine to the facts of this case.
Kaatz v. State,
The facts of this case are set forth at length in our prior opinion,
At the initial trial thе court, sitting without a jury, fixed the damages to Kaatz’ survivors at $719,000, but held their recovery was entirely barred by the rule of contributory negligence. On appeal, we held that contributory negligence would no longer be a total bar to recovery, but that under the rule of comparative negligence the recovery would be reduced in the proportion that the negligence of the plaintiff (or, here, the plaintiff’s decedent) contributed to the injury. We remanded the case so that the trial court might apportion the negligence between Kaatz and the state, and reduce the recovery accordingly.
We remanded fоr another purpose as well. The final paragraphs of our prior opinion read:
“We observe that in the present case defense counsel offered no evidence with regard to the issue of damages. This may have occurred because of a belief by defense counsel that under the contributory negligence doctrine plaintiffs’ claims should have been barred entirely. Since we have abandoned the contributory negligence doctrine today, we feel that it is appropriate to require a new trial on the issue of damages.
“We remand this case to the superior court for further proсeedings consistent with this opinion.”540 P.2d at 1050-51 .
After our mandate, the state, relying on the repeal of AS 09.50.290, 2 filed a demand for trial by jury on the damages issue. Kaatz moved for apportionment of negligence between the parties, after briefing, the motion came on for hearing on July 23, 1976. The court denied a jury trial, took no new evidence on damages, and amended its findings of fact and conclusions of law to reflect that, for unstated reasons, it apportioned fifteen percent of the negligence to Kaatz and eighty-five percent to the state. It reaffirmed its prior finding that $719,000 was the proper amount of damages, and enterеd judgment in favor of Kaatz for eighty-five percent of $719,000, or $611,150. The state appealed.
The state’s contentions on appeal are:
1. The trial court violated this court’s mandate, which ordered a new trial at which the state could produce evidence on the damages issue which it had not produced before;
2. The state was entitled to trial by jury on the damages issue;
3. Although comparative negligence has been adopted in Alaska, State v. I’Anson,529 P.2d 188 (Alaska 1974), holds that the state has no duty to keep its highways safe for negligent highway users;
4. The trial court’s findings of fact are inadequate for appellate review;
5. The fifteen to eighty-five percent apportionment is contrary to the evidence and clearly erroneous; and
6. The percentage apportionment should have been made by a jury.
I
The first issue requires reversal and remand. Our prior mandate directed the superior court to hold a new trial on the issue of damages, and receive new evidence on that issue. The superior court did not do so, although counsel for the state demanded a trial by jury and orally reminded the court of our instructions.
“[UJpon remand of a case by this court it becomes the duty of the lower court to obey the mandate and render judgment
*779
in conformity.”
State v. Salinas,
Although our decision on this issue is dispositive of the appeal, we will in this opinion address several of the other issues in the interest of judicial economy, since they will probably arise again in the superi- or court on remand.
II
AS 09.50.290 provided that tort claims against the state were to be tried by the court without a jury. It was repealed by section 1 of chapter 147, SLA 1975, which became effectively immediately upon signature by the governor, June 15, 1975. The first trial of this ease took place in 1973, when AS 09.50.290 was in effect, so jury trial was not requested and trial was to the court.
On September 29,1975, the day this court issued its mandate on the prior appeal, the state filed a written demand that the retrial on damages be by jury. Kaatz filed no papers mentioning the jury issue. At the July 23,1976, hearing, Judge Carlson denied a trial by jury. (As discussed in section I, supra, he did not conduct a trial at all.)
We must decide whether the repeal should be given limited retroactivity to this claim which arose and was filed long before the repeal, and was sub judice on appeal when the repeal took effect. Legislative history sheds no light on the question. The bill passed both chambers unanimоusly and without amendment. No substantive committee reports appear in the journals.
Statutes are presumed to operate prospectively only. AS 01.10.090. This does not answer the question, however, because the statute can be interpreted to mean either that jury trial was available only for claims arising or filed after its effective date, or that trials commencing after the statute took effect could be by jury.
A combination of several factors persuades us that the retrial of this case shall be by jury. The legislature gave the repeal immediate effect, waiving by two-thirds vote the normal rule that statutes take еffect ninety days after enactment. Alaska Const, art. II, § 18. Kaatz cannot claim a vested right in having trial only to the court.
Bidwell v. Scheele,
Ill
The previous appeal settled that both the state and Kaatz were negligent.
*780 The state bases this argument on I’An-son, supra at 195. In I’Anson, we rejected the state’s argument that it should be held to a lesser standard of care than private individuals. We then said:
“Thus, the appropriate standard of care required of [the state] and its agents was to use reasonable care to keep the highway in a safe condition for the reasonably prudent traveler.”
(Emphasis added.) A footnote at that point quoted
Martin v. State, Dep’t of Highways,
“ ‘Generally, it is the duty of the highway department to construct and maintain the highways . . . reasonably safe for a traveler himself exercising ordinary care; a highway will be deemed safe within these requirements if it may be negotiated successfully by all but the very reckless and careless drivers, there being no obligation to construct and maintain highways so as to insure the safety of such drivers.’ ”
The state now asserts, that before Kaatz I, its duty to negligent persons was merely an academic issue since a plaintiff who was negligent was barred from recovery for that reason alone. After Kaatz I, the state says, the plaintiff’s negligence without more eliminates the state’s duty to that plaintiff.
We rejected this argument in
State
v.
Guinn,
In
IAnson,
we held that the state must be treated like any other litigant for these purposes.
The state argues that finding a duty here would make the state an insurer of highway safety. The state, however, is not an insurer as it is only liable if it is negligent. Moreover, the more negligent the plaintiff’s conduct, the lower the percentage of damages which the plaintiff will be able to recover from the state.
6
If the plaintiff’s conduct is reckless, wilful, or criminal [true mens rea crimes, not regulatory traffic offenses], his percentage may approach 100, or he may as a matter of law and public policy be barred from recovery. Nor would the state be required to foresee wilful misconduct. Prosser,
supra,
at 173-74. The plaintiff in a tort action against the state must prove, as in any other tort action, that the defendant’s aсtion was both a factual and a legal cause of his injuries. If the causal connection is too tenuous, or if the plaintiff’s own conduct is a superseding cause, the plaintiff cannot recover even though the state owes him a duty of due care.
See generally, e. g., State v. Abbott,
Ikene v. Maruo,
Therefore, we hold that I'Anson must be read in light of our adoption of comparative negligence in Kaatz I, and that the dictum in I’Anson quoted above does not bar this action.
IV
The state makes three assignments of error with respect to the trial court’s apportionment of fifteen percent of the negligence to Kaatz and eighty-five percent to the state. The state argues that the court’s findings of fact are inadequate for appellate review of this point, that the apportionment is clearly erroneous, and that the apportionment should have been the subjeсt of a jury trial. We reject all three contentions.
The trial court concluded after our previous remand that fifteen percent of the negligence was attributable to Kaatz and eighty-five percent to the state. No other new findings or conclusions were made. At the end of the initial trial, before the first appeal, the court had made 23 fairly extensive factual findings, which were sustained on the previous appeal, with the exception of those relating to Lindley, who is no longer a party.
Civil Rule 52(a) requires findings of fact in cases tried without a jury, and we have made clear that they must be sufficient in quality and quantity to facilitаte meaningful appellate review. E. g., I’An-son, supra at 197 & n. 38 (Alaska 1974) (collecting other citations). As a general rule, “findings of fact must include as much of the subsidiary facts as is necessary to disclose to the reviewing court the steps by which the trial court reached its ultimate conclusion on each factual issue.” 9 C. Wright & A. Miller, Federal Practice and Procedure § 2579 at 710 (1971).
There is very little case law on what constitutes compliance with rules similar to our Rule 52(a) in comparative negligence cases. Most comparative negligence cases reach appellate courts from a judgment entered on a jury verdict, so written findings of fact are not required. This appeal is our first oсcasion to consider this question under comparative negligence principles, whether based on findings of fact by the court or a jury.
Initially, it is clear the the apportionment of percentages is a question of fact, not of law.
St. Hilaire Moye v. Henderson,
In
Hornsby v. Fishmeal Co.,
Comparing negligence, and assigning numerical percentages to the negligence of each party, is necessarily a discretionary process, as we discuss in somewhat more detail later in this opinion. Inherently it is extremely difficult for a trier of fact to describe why it chooses fifteen-eighty-five rather than ten-ninety or twenty-eighty. Juries are not called upon to explain the percentage figures they assign in their verdict, or to describe the process or the reasoning they used to reach them. The in *782 structions given to juries in comparative negligence cases frequently contain little or no specific guidance on how negligence is to be compared. See footnote 8, infra.
Findings of fact aid the trial court in adjudication by requiring it to articulate the factual premises of its decision, and also aid the appellate court by making those premises clear. See 5A Moore’s Federal Practice ¶ 52.06[1] (2d ed. 1975). Given the nature of the process of comparing negligence, we do not believe it would be reasonable to require the trial court to articulate the basis of its reasoning prоcess in any greater detail than it did in this case. Nor do we perceive that additional findings of fact would materially aid us in determining whether the percentage apportionment is clearly erroneous.
We emphasize that the findings of fact in this case are not limited to the bare percentages. The trial court made twenty-three separate findings of fact in addition to the percentage apportionment. They detail the events surrounding the death of Donald Kaatz, the ways in which each party was negligent, and the causal connection between the negligence and the death of Kaatz. Without these specific findings, the percentages alone would be inadequate for appellate review. Therefore, we reject the state’s suggestion that we either make supplementary findings of fact ourselves, or remand for further findings by the superior court. We hold that the findings of fact are adequate to meet the requirements of Rule 52(a).
We next turn to the question whether the finding that fifteen percent of the negligence was attributable to Kaatz and eighty-five percent to the state, is clearly erroneous. As a matter of law, the state says, at least fifty percent of the negligence should be apportioned to Kаatz. In considering this question, we will briefly and generally discuss the standards which the trier of fact should use in apportioning negligence.
At the outset, we recall the standard we use in reviewing findings of fact. Civil Rule 52(a) states that:
“Findings of fact shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses.”
A finding is clearly erroneous only if “we are left with the definite and firm conviction on the entire record that a mistake has been committed.”
Alaska Foods, Inc.
v.
American Mfrs. Mutual Ins. Co.,
What is to be compared is negligence — conduct, fault, culpability — not causation, either physical or legal.
Pan-Alaska Fisheries, Inc.
v.
Marine Const. & Design Co.,
Briefly summarized, the facts indicating Kaatz’ negligence are these: The Mitkof Highway was icy and very slippery, which condition was apparent to the casual observer. Kaatz had driven the section of the Mitkof Highway in question shortly before riding in the loadеr. Loaders are not designed for travel on a highway, but they are often driven over highways under their own power. Kaatz was familiar with the characteristics of the loader. He also knew the condition of the road. The loader is designed to carry only a driver, not a passenger, and the policy of Kaatz’ employer forbade riding on loaders, with exceptions not relevant to this case. Kaatz was not the first person to ride as a passenger on a *783 loader in violation of policy. 7 On previous trips, Kaatz had ridden in a pickup truck instead of on the loader. In those circumstances, Kaatz was negligent in riding in the loader on the Mitkof Highway.
The facts frоm which the state was held negligent in brief, are: The state highway department had a standard operating procedure that called for sanding of roads in inclement weather, even in outlying areas and even if it required overtime or night work. Cecil Donohue, the state maintenance foreman in the Petersburg sector, knew of the slippery condition of the Mit-kof Highway before the accident occurred. His sanding machine did not work, but repairs were being undertaken. His stockpile of sand was frozen and covered with snow. He drove the stretch of road in question a few hours before the accident but decided not to sand it until the next morning, because he thought it might snow and cover the sand, even though he thought the road probably needed sanding that evening. Later that evening, after the accident, he and his men sanded the road. In those circumstances, the state was negligent in failing to sand the road in the late afternoon.
The state’s foreman testified that the road needed sanding for regular traffic, not for front-end loaders. Two nearby residents testified the road was worse that evening than they had ever seen it before. If he had considered only the testimony of Mr. Schouweiler, one of these nearby residents, the trial judge said he would have found Kaatz at least fifty percent negligent. On the other hand, he pointed out, other witnesses testified that they had negotiated the highway without inordinate difficulty. Kaatz himself drove the highway in a pickup truck without apparent incident shortly before riding the loader. Kaatz suggests another aggravating factor concerning the state’s negligence: it exposed large numbers of persons — everyone who traveled the highway — to danger, whereas Kaatz’ negligence had the potential to harm no one but himself.
That Kaatz was acting in the course of his employment mitigates his negligence somewhat. In
Bachner
v.
Rich,
The state asserts that Kaatz was actively negligent and the state passively negligent, and therefore as a matter of law the greater percentage of the negligence must be assigned to Kaatz. The state takes the active/passive concept from factual contexts different from this one. It is not clear that the state’s negligence should be classified as “passive” and Kaatz’ as “active.” In any event, we see little to be gained in importing the active/passive distinction into the comparison of negligence. One of the virtues of comparative negligence is its greater flexibility. See Schwartz, supra, § 21.2 at 340. Introducing various standards and concepts from other areas of tort law, and creating from them rigid rules to be used in comparing negligence, would destroy much of that flexibility.
Essentially the same reasoning persuades us to reject the state’s reliancе on
Rewolinski v. Harley-Davidson Motor Co.,
We cannot offer specific guidelines on how to compare negligence. Every case must turn on its own facts. The trier of fact, whether judge or jury, must аpply its ordinary human experience to the facts revealed by the evidence. The pattern jury instructions used in Wisconsin, perhaps the leading state in the development of comparative negligence law, and in California offer no real guidance. 8 See also Aiken, Proportioning Comparative Negligence-Problems of Theory and Special Verdict Formation, 53 Marquette L.Rev. 293, 294-97 (1970).
Professor Victor Schwartz, in his treatise on comparative negligence, offers four items which make the apportionment process somewhat more concrete for the trier of fact. He suggests that counsel might find them helpful in structuring the closing argumеnt. They are: (1) the probability from each negligent party’s point of view that the particular kind of harm that actually occurred, would indeed result; (2) the extent to which reasonable people in the positions of the parties could foresee that the particular harm might occur; (3) what, if anything, was to be gained by each party taking the risk it did; and (4) if something of significant value was to be gained, whether there were more reasonable alternatives open to the party. Schwartz, supra, § 17.1 at 278. 9
If we were the triers of fact, we might well have apportioned the negligence more favorably to the state than did the trial court. But aftеr examining the entire record, we are not left with the definite and firm conviction that the trial court was clearly mistaken. 10 Therefore, we affirm *785 its finding apportioning the negligence fifteen percent to Kaatz and eighty-five percent to the state.
The state’s final argument for vacating the apportionment of negligence is that the superior court judge should have submitted the question to a jury, instead of making an apportionment himself based on the evidence received at the initial trial before the first appeal.
In our opinion on the prior appeal, we did not indicate whether the apportionment should be made by the judgе on the basis of the record of the initial trial, or whether a new trial should be held on that issue. In our more recent opinion in
State v. Guinn,
The state argues that, since the case is being remanded for a new trial on damages, the burden on the superior court and counsel would be slight if the apportionment issue were also remanded for a new trial. 11 We disagree. Apportionment of negligence is part of the liability phase of a case, not the damages phase. If the trier of fact performs its function correctly, it will consider the two issues wholly separately. See, e. g., California Jury Instructions — Civil, supra, 14.90.
The jury which apportions the negligence must hear all the evidence on liability. In the instant case, the evidеnce on liability is almost entirely separate from the evidence on damages. Kaatz was apparently killed instantly. No survival damages or other damages directly relating to the events of December 9, 1970, are sought. At the retrial of the issue of damages, the evidence will concern Kaatz’ future earning capacity, his life expectancy, his habits of consumption, and so forth.
Were the jury also to decide the apportionment of negligence, the parties would be required, in addition, to present all the evidence on the events of December 9,1970, which was presented at the initial trial. We do not find that cоnsiderations of judicial economy would be served by a retrial of the apportionment of negligence.
V
In summary, we reverse the judgment appealed from, and remand the case for a new jury trial on the issue of damages. Following the trial, the court shall, in entering judgment, reduce the sum of damages found by the jury by fifteen percent, to *786 reflect the apportionment of fifteen percent of the negligence to the decedent, which apportionment we affirm.
AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.
Notes
. Kaatz’s co-worker, Ronald Lindley, was also killed in the incident. Prior proceedings in this case have been consolidated with the wrongful death action brought by Lindlеy’s survivors. That action has been settled, and is not before us on this appeal.
. AS 09.50.290 provided:
“Actions against the state under §§ 250-300 of this chapter shall be tried by the court without a jury.”
AS 09.50.290 was repealed by § 1 ch. 147 SLA 175, effective June 15, 1975.
.
See
Alaska Const. art. I, § 16.
Compare, e. g., Loomis Electronic Protection, Inc. v. Schaefer,
. Most such highway users will' probably be drivers. Kaatz was a passenger, but that is irrelevant, particularly as his negligence related in part to his having driven the highway in question a short time earlier.
See
. We said:
“The State cannot be permitted to rely upon the sharpened skills and good fortune оf passing motorists to discharge its duty of due care to the driver exercising ordinary care.”
Id. at 538 (emphasis added). Despite this reference to non-negligent drivers, we permitted the plaintiffs in Guinn to recover although their decedent himself had been negligent.
. Much of the state’s argument that it owes no duty to negligent highway users is actually a claim that Kaatz ought to have been assigned a very high percentage of the negligence, and the state very little. The latter assertion we take up further in this opinion.
. At the end of the original trial, Judge Carlson stated it was impossible to determine which man was the driver. There appears to be no dispute at the present stage of the proceedings that Lindley was the driver and Kaatz the passenger.
.The Wisconsin instruction, as reprinted in C. R. Heft and C. J. Heft, Comparative Negligence Manual § 7.520 (Supp.1976), reads as follows:
“If you answer Question-[or this question] you will determine how much or to what extent each person (party) named in the question is to blame for the (collision) (accident! (injury) and considering the conduct of the persons (parties) named in the question, considered as a whole, you will determine whether one made a larger, equal or smaller contribution than the other (others). In making your apportionment of negligence you will fix the percentage attributable to each participant in proportion to how much the fault of each contributed to cause the (collision) (accident) (injury) and record those percentages.”
The California instruction, California Jury Instructions — Civil [BAJI] 14.91 (Supp.1975), entitled “Guidelines in Determining Comparative Negligence,” reads:
“In order to determine the proportionate share of the negligence attributable to the plaintiff, you will of necessity be required to evaluate the combined negligence of the plaintiff and of the defendants] [and of all other persons] whose negligence proximately contributed to plaintiff’s injury.
“In making this evaluation, you are instructed that it is the negligence of such persons that you must measure and not the mere physiсal causation for the accident.
“In comparing the negligence of such persons you should consider all the surrounding circumstances as shown by the evidence.”
. In
Associated Engineers, Inc. v. Job,
. We bear in mind the advice of the Supreme Court of Wisconsin in the first case to reach it after enactment of that state’s comparative negligence statute:
“When two persons are negligent and injury to one proximately results from the combined negligence of both, it must often be a very delicate and difficult question to decide whether the negligence of one was greater than that of the other, and contributed in a greater degree tо produce the injury. There is no yardstick with which to measure the two acts of negligence, nor scales with which to weigh them.”
*785
Brown
v.
Haertel,
. New trials on the sole issue of percentage apportionment are recognized in some other comparative negligence states, including Wisconsin, but they are seldom granted because they require presentation of all the evidence on liability, unlike most new trials limited to the issue of damages.
See
Schwartz,
supra,
§ 18.3 at 301-05. If a new trial on the issue of apportionment is granted, the court should instruct the jurors that (1) they must assign some percentage of the negligence to each party, (2) the total damages are in the sum found by the previous trier of fact (unless the same jury is also retrying the damages issue), and (3) their sole function is to apportion the negligence between the parties.
See, e. g., O’Kelly v. Willig Freight Lines,