CALABRESI, Circuit Judge:

I. BACKGROUND

On May 28, 1989, Marvin Eichelberg and his brother, James, decided to go fishing. They left their car at a marina in Noank, Connecticut, and, bypassing a fence, walked onto railroad tracks owned and operated by Amtrak. They followed the tracks to the point at which Beebe Cove met Long Island Sound. There, a trestle, approximately 14 feet in length, spanned the entry to the cove. Deciding that the south end of the trestle looked like a promising spot, the Eichelberg brothers began to fish. They were four to five feet from the railroad tracks.

A half hour's fishing failed to produce a catch. The brothers therefore thought to try their luck at the north end of the trestle. They went along the trestle until they reached the abutment at the north end, where they remained for another fifteen or twenty minutes. Still the fish refused to bite.

Eichelberg then decided to try another strategy. Climbing back onto the trestle, he cast his line and proceeded to walk slowly to the center of the trestle, casting, reeling in the line, and casting again as he went. Eichelberg later testified in his deposition that he knew trains ran over the tracks from which he was fishing, but that he believed he was exercising appropriate care:

I considered it to be safe in much the same way I would cross a busy street in the city; throwing caution over my shoulder and proceeding would be idiotic. I didn't do that. I watched for the train figuring I had a darn good — I could see for a mile in one direction and could see for hundreds of feet in the other, and with what you would say with calculated caution, I considered what I was doing as safe procedure.

Once he reached the center of the trestle, Eichelberg cast his line two or three times, and began slowly to make his way back toward the north end of the trestle, all the while continuing to fish. Eichelberg had only proceeded four or five feet to the north when he spotted a train approaching from the south.

George Craig, the engineer aboard the northbound Amtrak train, testified in his deposition that the train was traveling at about 60 to 65 miles per hour, under a "clear" signal, when it rounded a bend in the tracks and came within view of the trestle from which Eichelberg was fishing. When Craig spotted Eichelberg, he sounded one long blast on the train's whistle. Craig did not, however, immediately apply the brakes. He testified that it was Amtrak's policy for the engineer to blow the whistle whenever someone was on or very near railroad tracks when a train was approaching. He further testified that people usually had the presence of mind to get off the tracks when confronted with an oncoming locomotive, and that brakes generally would not be applied unless it appeared that the train would hit the trespasser on the tracks.

Eichelberg's and Craig's accounts of what happened next differ in some respects. Eichelberg testified that he saw the train even before it first sounded its whistle, and that he immediately began moving as quickly as he reasonably could toward the north end of the trestle, at a pace that he described as "a slow, calculated trot or an extremely fast walk." Eichelberg's progress was allegedly slowed by the need to ensure that his steps fell on the railroad ties that connected the tracks, and not into the gaps between the ties. Eichelberg undoubtedly was also slowed by the fact that he continued to keep hold of the fishing pole; indeed, although Eichelberg denied that he continued to fish once he saw the train, he testified that he may have been reeling in his line as he made his way to the northern end of the trestle. In Eichelberg's recollection, he heard three additional blasts of the train whistle just as he reached the northern end of the trestle and was preparing to jump from the tracks.

Besides making his way to the northern end of the trestle and leaving the tracks once he got there, Eichelberg had two possible options to avoid the oncoming train. First, he might have jumped from the trestle. Eichelberg testified, however, that this was not a realistic option. It was low tide, he explained, and instead of simply jumping into water he would have fallen a good distance onto rocks. Second, since the trestle had tracks running in each direction, separated by a narrow mezzanine, Eichelberg might have moved to the track that the train was not using, or at least to the mezzanine. Eichelberg testified that he did not consider these to be viable choices, either. He said that he did not know which track the train was on, because it was approaching on a curve. He added that it would have taken him some time to cross the tracks and get to the mezzanine area. And he stated that he also believed that there was no place on the trestle where he could stand safely while the train passed by (a belief supported by the deposition testimony of the train's engineer). In Eichelberg's judgment, therefore, it was better to try to get off the track at the northern end of the trestle.

Craig, in his deposition, described a rather different series of events. Craig testified that when he first blew the whistle, Eichelberg looked up at the train and began "moving very slowly, . . . sauntering off of this bridge." Craig, becoming concerned, gave a series of blasts on the horn. Eichelberg responded with what Craig described as "a gesture of impatience" and continued moving slowly off the trestle. At this point, when he was approximately 600 or 700 feet from the bridge, Craig realized that Eichelberg was making no serious effort to get out of the way, and so he allegedly applied the emergency brake. As the train approached Eichelberg, Craig saw that Eichelberg was "reeling his line in and seemed more concerned with his fishing than he did with getting hit by a train." According to Craig, he applied the emergency brake as soon as he realized that Eichelberg was not making a serious effort to get off the trestle and thus was in a perilous situation.

Eichelberg's brother, James, provided a third perspective on the day's events. James testified that he initially heard the train sound its whistle when it was some distance away, and that it did not sound the whistle again until it was about 50 to 55 feet from the southern end of the trestle. James also testified that the train did not appear to be slowing down as it crossed the trestle, and that at no time did he hear the squeal of train brakes or observe sparks coming from the rails.

Although there is disagreement about the events leading up to impact, there appears to be no dispute about what happened next. Just as Eichelberg prepared to jump from the abutment at the end of the trestle, he was struck in the area of his left shoulder blade by a grab handle that protruded from the side of the train's engine car. Eichelberg was thrown into the air by the impact and landed on the rocks that lay below the abutment. Eichelberg alleges that, as a result of the accident, his left arm was partially severed. Although doctors were able to reattach it, his use of the arm is now limited, and he suffers persistent numbness and pain.

Eichelberg filed suit against Amtrak in Connecticut Superior Court, claiming that his injuries were caused by Amtrak's negligence. Amtrak, as a corporation created by Congress and owned by the federal government, removed the action to federal court pursuant to 28 U.S.C. § 1441. Following discovery, Amtrak moved for summary judgment, which the district court granted. This appeal followed.

II. DISCUSSION

A. A landowner's duty to trespassers

Whether summary judgment is appropriate in this case turns in significant part on the nature of the duty that Amtrak, as an owner and occupier of land, owed to Eichelberg, a trespasser upon that land. Unlike many jurisdictions, which have abolished some or all of the common-law distinctions based on the status of individuals who enter onto land, see generally W. Page Keeton et al., Prosser and Keeton on The Law of Torts § 62, at 432-33 (5th ed. 1984), Connecticut, whose law we apply, continues to observe "an ascending degree of duty owed by the possessor of land to persons on the land based on their entrant status, i.e., trespasser, licensee or invitee." Morin v. Bell Ct. Condominium Ass'n, 223 Conn. 323, 327, 612 A.2d 1197, 1199 (1992). As to trespassers, the general rule is that a landowner's duty is limited to the duty not to injure the trespasser intentionally or by willful, wanton, or reckless conduct. See Morin, 223 Conn. at 328, 612 A.2d at 1200; McPheters v. Loomis, 125 Conn. 526, 531, 7 A.2d 437, 440 (1939); Kalmich v. White, 95 Conn. 568, 571, 111 A. 845 (1920); Whitney v. New York, New Haven Hartford R.R. Co., 87 Conn. 623, 633, 89 A. 269 (1914).

Connecticut, however, recognizes several important exceptions to this general rule. The district court addressed two of these. Under the first exception, if the land-owner "know[s] that the presence of trespassers is to be expected, then the common obligation of exercising reasonable care gives rise to the correlative duty of taking such precautions against injuring trespassers as a reasonable foresight of harm ought to suggest." Carlson v. Connecticut Co., 95 Conn. 724, 730, 112 A. 646 (1921); see also Morin, 223 Conn. at 333, 612 A.2d at 1202.

The district court correctly concluded that Eichelberg cannot avail himself of this exception. To demonstrate that Amtrak had knowledge of trespassers on or near the trestle sufficient to trigger a duty of care, Eichelberg must show that the presence of trespassers was "reasonably regular and predictable." Morin, 223 Conn. at 331, 612 A.2d at 1201. Eichelberg introduced evidence that there had been one previous collision, in 1980, between a trespasser and an Amtrak train at the trestle. Eichelberg also introduced evidence that Amtrak was aware that people occasionally fished in the area of the trestle, although not from the trestle itself. We believe that this evidence falls short of demonstrating that the presence of individuals fishing on or near the trestle was "reasonably regular and predictable" enough to trigger the duty of care.

Similarly, Eichelberg may not rely on evidence that Amtrak was aware that trespassers were commonly seen at various different spots along its tracks. To trigger the duty of care under the Carlson exception, Eichelberg must show that Amtrak had notice, not merely that a trespasser might reasonably be expected to be somewhere along its tracks, but that Amtrak had reason to believe that trespassers would be near the particular place where the accident occurred, or at another place with similar characteristics. In Carlson, for example, the defendant had clear notice that on Saturday evenings drunken men were frequently found on or near a particular stretch of trolley tracks that ran along a highway. The court, therefore, found that the defendant owed a duty of care to the trespassing inebriates at that time and place. Carlson, 95 Conn. at 726, 730, 112 A. 646. Nothing in Eichelberg's proof even remotely approaches the precision of the notice that the Carlson defendant possessed. Having failed to introduce evidence sufficient to support a finding that Amtrak knew that trespassers commonly fished on or near this trestle or similar trestles, Eichelberg cannot meet the requirements of the Carlson exception.

Under the second exception recognized by the district court, a landowner is liable for injuries to children caused by an artificial condition on the landowner's property if the landowner knows or has reason to know that children are likely to trespass and are not likely to appreciate the danger. Wolfe v. Rehbein, 123 Conn. 110, 113-14, 193 A. 608 (1937); see also Morin, 223 Conn. at 333, 612 A.2d at 1202; Duggan v. Esposito, 178 Conn. 156, 158, 422 A.2d 287, 290 (1979). As the district court observed, Eichelberg, who at the time of the incident in question was a 34-year-old adult of supposedly sound mind, cannot avail himself of the Wolfe exception.

Were the two doctrines mentioned by the Connecticut Supreme Court in Morin and relied upon by the district court the only exceptions to the Connecticut rule that a landowner owes no duty of ordinary care to trespassers, we would affirm. The Morin court's discussion of these two exceptions to the general rule did not purport to be exhaustive, however. See, e.g., Morin, 223 Conn. at 332, 612 A.2d 1197 (stating that Carlson and Wolfe set forth "two of the exceptions to this general rule"). And Connecticut courts have long recognized a third exception, holding that "when the presence of a trespasser in a position of peril becomes known, the duty then arises of using ordinary care to avoid injuring him." Kalmich, 95 Conn. at 571, 111 A. at 845; see also Kuharski v. Somers Motor Lines, 134 Conn. 35, 42, 54 A.2d 592, 595 (1947); Kuharski v. Somers Motor Lines, 132 Conn. 269, 273, 43 A.2d 777, 778 (1945); Shiembob v. Ringling, 115 Conn. 62, 65, 160 A. 429 (1932); Hess v. Springfield Terminal Ry. Co., 1994 WL 669551, at *2 (Conn.Super.Ct. Nov. 21, 1994).

A trespasser is "in a position of peril" when the trespasser is "in a situation where there is a danger he may be injured if the defendant fails to exercise reasonable care under the circumstances." Kakluskas, 134 Conn. at 42, 54 A.2d 592. Given this broad definition of "position of peril," it is possible that Eichelberg was in a position of peril as soon as he came within view of the train. This does not necessarily mean, however, that Craig was required, as Eichelberg contends, to apply the brakes as soon as he saw Eichelberg on the tracks, because Connecticut law also imposes on Eichelberg a duty to avoid the danger in which he found himself. Sitnik v. National Propane Corp., 151 Conn. 62, 65, 193 A.2d 503, 505 (1963). As the Restatement puts it,

[t]he possessor of land is entitled to expect that a trespasser who is warned or otherwise has reason to believe that the possessor intends to do an act which involves danger to the trespasser will yield precedence to him and, knowing of the danger, will avoid it. Therefore, the possessor may, if he gives reasonable warning, continue his activities in such a way as he pleases unless the situation is such that he should realize that a warning will not be effective to enable the trespasser to avoid harm.

Restatement (Second) of Torts § 336, cmt. d(1) (1965). Thus, under ordinary circumstances, when a railroad engineer sees a pedestrian on the tracks some distance away, the engineer is entitled to assume, at least at first, that, given an appropriate warning such as a blast from the whistle, the pedestrian will take the (usually minimal) precautions to avoid the obvious danger of the onrushing train. See, e.g., Box v. South Georgia Ry. Co., 433 F.2d 89, 93 (5th Cir. 1970); Maxwell v. Illinois Cent. Gulf R.R., 513 So.2d 901, 905 (Miss. 1987).

Eichelberg's situation may not have been ordinary, however. Unlike the pedestrians in Box and Maxwell, Eichelberg was not on tracks that ran along dry ground, which obviously would have allowed him to move to safety in just a few steps. Cf. Box, 433 F.2d at 93 (the defendant was entitled to presume that the plaintiff "would take the necessary two or three steps to place her in a position of safety"). Rather, he was in the middle of a relatively narrow trestle, which at low tide spanned a rocky inlet. A jury could find that jumping from the trestle was not a safe option, and Eichelberg introduced evidence that moving to the other track on the trestle or to the mezzanine between the two tracks was not a viable choice, either.

More important, under Connecticut law, once the situation was such that it should reasonably have become evident to Craig that Eichelberg either would not or could not move to safety, merely relying on the whistle was no longer by itself sufficient to meet the duty of care imposed on Craig, and on Amtrak as Craig's employer. See, e.g., Marchand v. New York, New Haven Hartford R.R. Co., 146 Conn. 599, 605, 153 A.2d 438, 441 (1959) (the duty of care imposed on a railroad engineer who knows an individual is on the railroad tracks may require, in appropriate circumstances, that the engineer reduce the speed of the train). See also Restatement (Second) of Torts § 336 cmt. d(1), illus. 3 (1965) (where an individual is walking along railroad tracks and the engineer, after blowing the train's whistle, sees that the individual will not obey the warning, the engineer must take reasonable care to stop the train so as to avoid hitting the pedestrian). And Craig essentially admitted that he came to recognize that Eichelberg was in a position of peril as the train approached the trestle when he realized "that this guy didn't intend to get out of the way" and that it was "starting to look like [he was] not going to get off the bridge."

Craig testified that he applied the emergency brakes as soon as he realized that Eichelberg was in a position of peril, when the train was, he estimated, about 600 or 700 feet from the trestle. Unless Craig should have recognized the danger to Eichelberg earlier, Craig's deposition testimony would certainly support a conclusion that he exercised reasonable care under the circumstances at that point. See, e.g., Goudreau v. Connecticut Co., 84 Conn. 406, 409-10, 80 A. 281 (1911) (a trolley operator who applied the brakes as soon as he saw the decedent start to run toward the trolley was not negligent).

That testimony is not uncontradicted, however. Eichelberg's brother, James, testified that the train did not appear to be slowing as it crossed the trestle, and that he neither heard the squealing of brakes nor saw sparks flying from the rails. From this testimony a jury might reasonably conclude that Craig had failed to apply the emergency brakes even after he had realized that Eichelberg was in a position of peril. Of course, the jury would not be bound to believe James's testimony or to draw the conclusion suggested here. But credibility determinations are within the province of the jury and may not be resolved on a motion for summary judgment. See United States v. Rem, 38 F.3d 634, 644 (2d Cir. 1994).

A material question of fact therefore existed as to whether Craig breached the duty that he and Amtrak owed to Eichelberg under the Connecticut "position of peril" exception to the general rule of no duty of ordinary care to trespassers. Summary judgment, then, was inappropriate unless, for some reason, Connecticut would no longer recognize the exception or unless Eichelberg was so negligent as to be barred from recovery, as a matter of law, regardless of the exception. Neither of these questions is as simple as may at first appear.

B. The continued vitality of the "position of peril" exception

The Connecticut Supreme Court has not mentioned the "position of peril" exception since 1947, see Kakluskas, 134 Conn. at 42, 54 A.2d at 595, although a trial court has recognized it as recently as last year. See Hess, 1994 WL 669551, at *2. And in the interim Connecticut tort law has undergone a dramatic upheaval, with the replacement of the common-law doctrine of contributory negligence by a statutorily-established scheme of comparative negligence. See Conn.Gen.Stat. § 52-572h.

The comparative negligence statute does not directly address the "position of peril" exception. The exception, however, bears an uncanny resemblance to the Connecticut version of the doctrine of "last clear chance," a doctrine that was explicitly abrogated by the statute. Conn.Gen.Stat. § 52-572h( l). The rule of "last clear chance" permitted a plaintiff to avoid the consequences of contributory negligence if the defendant, through the exercise of reasonable care, had the last clear chance to avoid the injury. The rule applied where:

(1) [t]he injured party, by his own negligence, has already come into a position of peril; (2) the injuring party then or thereafter becomes, or in the exercise of ordinary prudence ought to have become, aware not only of that fact but also that the party in peril either reasonably cannot escape from it or apparently will not avail himself of opportunities open to him for doing so; (3) the injuring party subsequently has the opportunity by the exercise of reasonable care to save the other from harm; and (4) he fails to exercise such care.

Childs v. Blesso, 260 A.2d 582, 583, 260 A.2d 582, 583 (1969); see also DePaola v. Seamour, 163 Conn. 246, 248, 303 A.2d 737, 739 (1972); Intelisano v. Greenwell, 155 Conn. 436, 444, 232 A.2d 490, 494 (1967); Fine v. Connecticut Co., 92 Conn. 626, 631, 103 A. 901 (1918). As is readily apparent, Connecticut's last clear chance doctrine allowed negligent plaintiffs to recover against negligent defendants in circumstances virtually identical to those in which Connecticut's "position of peril" exception permitted trespassers to recover against negligent landowners.

Does the abolition of last clear chance, therefore, mean that the "position of peril" exception has also been eliminated? We think not. The abolition of last clear chance in Connecticut, as in other states, occurred as part of the process in which the rule of contributory negligence was replaced by comparative negligence. See Conn.Gen. Stat. § 52-572h( l). That is, once a plaintiff's negligence was no longer an automatic bar to recovery — as it had been under the contributory negligence rule — then the last clear chance doctrine, which when it applied shifted the whole loss to the defendant despite the plaintiff's negligence, seemed both unduly harsh on the defendant and no longer necessary to alleviate the burden on those plaintiffs whose negligence was only partially responsible for the harm. Accordingly, last clear chance as a separate doctrine was done away with (though as we shall soon see some of its underlying elements have been incorporated into the judgment of relative responsibility required by comparative negligence law).

It may well be that had Connecticut done away with the rule that put on trespassers the whole burden of injuries that occurred while they were improperly on the land, and replaced this doctrine with one that considered trespass as one factor to be weighed against others in deciding liability (as other states have done), we would find that the "position of peril exception" to the trespasser rule had been abrogated. Placing the full loss on the landowner, merely because the landowner saw and should have realized that the trespasser was in peril, would make little sense once the fact of the trespass was no longer enough, by itself, to cause the loss otherwise to be borne solely by the plaintiff. The "position of peril" exception would, under such circumstances, seem as anachronistic as last clear chance seemed to be once contributory negligence was replaced by comparative negligence.

The Connecticut rule limiting a landowner's duty to trespassers perdures, however. And, under the circumstances, we believe that the "position of peril" exception survived the elimination of last clear chance. To find otherwise, we would have to conclude that the statute establishing comparative negligence, which was intended to do away with "the harsh rule that contributory negligence was a complete defense in negligence cases," Gomeau v. Forrest, 176 Conn. 523, 525, 409 A.2d 1006, 1007 (1979), and thereby to expand the possibility of a plaintiff's recovery (albeit perhaps at an amount less than the plaintiff's full damages), had the effect of eliminating any possibility of recovery in a series of cases in which, under the common-law rules, plaintiffs were entitled to recover their full losses. In the absence of either a clear indication of legislative intent or any indication from the Connecticut courts that such a result is warranted, we decline to read the liability-expanding comparative negligence statute in such a liability-restricting way.

C. Comparative negligence

To say that the "position of peril" exception survives in Connecticut, and that Eichelberg has raised sufficient issues of fact so that a jury could find that he is within that exception, does not end the matter, however. Connecticut's comparative negligence statute allows a plaintiff to recover only if the negligence of the defendants was greater than or equal to the plaintiff's own negligence. Conn.Gen.Stat. § 52-572h(b). This means that Eichelberg cannot prevail if his negligence exceeded Amtrak's. And if it were clear that Eichelberg's negligence exceeded Amtrak's as a matter of law, then summary judgment for Amtrak would be appropriate despite the applicability of the "position of peril" exception.

On the record before us, we have no doubt that Eichelberg was negligent. Eichelberg's protestations to the contrary notwithstanding, an adult of sound mind who chooses to fish from the middle of a railroad trestle which rises over rocks and shallow waters, knowing that trains traverse the trestle, knowing too that, because of the construction of the trestle, footing is uncertain and a hasty retreat will be difficult if not impossible, and not knowing which track is for northbound trains and which is for southbound trains, cannot be said to be exercising reasonable care.

But did Eichelberg's negligence exceed Amtrak's as a matter of law? The resolution of this question turns in part, oddly enough, on whether Eichelberg or Craig had a last clear chance to avoid the injury. For although the last clear chance doctrine has been abrogated in Connecticut, the usual practice — from which there is no indication that Connecticut would depart — is that the factors pertinent to the last clear chance doctrine remain relevant in assessing the comparative fault of the plaintiff and the defendant. See, e.g., Macon v. Seaward Constr. Co., 555 F.2d 1, 2 (1st Cir. 1977) (applying New Hampshire law) ("While the advent of comparative negligence may have set aside the strict last clear chance doctrine, it has uniformly been held that the components of the doctrine remain as proper factors for the jury to consider in apportioning fault."); Kaatz v. Alaska, 540 P.2d 1037, 1050 n. 32 (Alaska 1975) (although the doctrine of last clear chance is abolished with the adoption of comparative negligence, "[t]hat is not to say that the notion of last clear chance is unavailable as a matter of trial court advocacy"); Li v. Yellow Cab Co., 13 Cal.3d 804, 826, 532 P.2d 1226, 1242, 119 Cal.Rptr. 858, 873 (1975) ("last clear chance . . . [is] to be subsumed under the general process of assessing liability in proportion to fault"); Cushman v. Perkins, 245 A.2d 846, 850-51 (Me. 1968) (under a comparative negligence system, the components of last clear chance "remain as factors to be considered by the jury in measuring and comparing the parties' relative fault"); McIntyre v. Balentine, 833 S.W.2d 52, 57 (Tenn. 1992) (with the adoption of comparative negligence, "[t]he circumstances formerly taken into account by [the doctrine of last clear chance] will henceforth be addressed when assessing relative degrees of fault"); Dixon v. Stewart, 658 P.2d 591, 698-99 n. 7 (Utah 1982) (with the abolition of the strict last clear chance doctrine, "`last clear chance' becomes just one of many factors to be weighed in the comparison by the trier of fact"); cf. Wendland v. Ridgefield Constr. Servs., 190 Conn. 791, 797-98, 462 A.2d 1043, 1047 (1983) (factors relevant to assumption of risk may be considered in determining comparative negligence despite the abolition of the assumption of risk doctrine).

The continued relevance of the factors pertinent to last clear chance weighs strongly against entering summary judgment for a defendant in a case where the defendant's negligence was, or could be found to be, last in time. There are two reasons for this. First, before the passage of comparative negligence laws, plaintiffs would have recovered fully whenever their contributory negligence was accompanied by a defendant's last clear chance (at least in those jurisdictions, like Connecticut, where last clear chance applied). There may therefore be an understandable reluctance to read into the passage of comparative negligence, and the concomitant abolition of last clear chance, the intent to bar such previously fully compensated plaintiffs from any recovery whatsoever as a matter of law. Second, where the defendant was the last wrongdoer, it is difficult to say, clearly enough to justify a decision as a matter of law, that a plaintiff was nonetheless more responsible. That should, instead, be a jury question. Cf. Fleming James, Jr., Connecticut's Comparative Negligence Statute: An Analysis of Some Problems, 6 Conn.L.Rev. 207, 213 (1974) (stating that the negligence of a defendant who fails to take advantage of a last clear chance is usually at least as great as the plaintiff's prior negligence). Perhaps for these reasons, no Connecticut court of which we are aware has found summary judgment appropriate, even against a highly negligent plaintiff, in a situation where the defendant's negligence was last in time.

Like some other states, however, Connecticut did not restrict its last clear chance doctrine to situations in which the defendant's negligence was, in fact, last in time. Connecticut's version of the doctrine permitted recovery where a defendant ought to have realized that the plaintiff "[would] not avail himself of the opportunities open to him for escaping" from a perilous position, and the defendant thereafter failed to use reasonable care to take advantage of an opportunity to avoid the harm. Childs, 158 Conn. at 392, 260 A.2d at 583. This approach meant that a plaintiff whose negligence was last in time, and who therefore, strictly speaking, had the last chance to avoid injury (that is, who could have prevented the accident through reasonable care even after the defendant could not), nevertheless could prevail against a defendant who had what might be termed the "clearer last chance" (that is, a defendant who should have realized that the plaintiff would continue to behave carelessly and who did not act reasonably to protect the plaintiff).

Some of the situations in which "clearer last chance" has come up are similar to the one before us. A classic case involved an elderly and disabled man who, riding a tricycle, attempted to cross a street in front of a trolley car. The trolley car rang its bell. The driver should have realized that the elderly man (who, it turned out, was hard of hearing) was not paying attention to the bell. Instead of braking, however, the driver kept ringing the bell. At a certain point the driver could no longer have avoided a collision by braking. But the elderly man could still have turned the tricycle away from the trolley. Had he simply bothered to look — as was his duty — he could easily have escaped harm. He did not do so, and yet under this "clearer last chance" version of the last clear chance doctrine, he was permitted to recover. See Locke v. Puget Sound Int'l Ry. Power Co., 100 Wn. 432, 171 P. 242 (1918).

The fact that Connecticut's version of the last clear chance doctrine allowed recovery for some plaintiffs whose negligence was in fact last in time has implications for comparative negligence cases in Connecticut. Even though we are confident that summary judgment would not lie against plaintiffs injured by defendants whose negligence was last in time, we cannot be equally confident as to what Connecticut would do when — although the situation is one in which Connecticut's last clear chance doctrine would have applied — it is the plaintiff who actually had the last chance to avoid the injury by acting reasonably. Earlier, we gave two reasons for the apparent judicial reluctance to grant summary judgments against plaintiffs in cases where, before the advent of comparative negligence, last clear chance would have applied: (1) the notion that a plaintiff who would have won before the coming of comparative negligence ought at least to be entitled to a jury trial after the enactment of that allegedly "pro-plaintiff" change, and (2) the belief that the last wrongdoer — i.e., the party whose negligence is last in time — cannot be deemed by a court to be less responsible as a matter of law than a party whose negligence has ended. Only the first of these applies when the plaintiffs negligence is last in time. Without any guidance from the Connecticut courts, we hesitate to decide how much weight Connecticut would give to the fact that a defendant had the clearer last chance now that comparative negligence reigns and last clear chance has been abolished.

It is not clear, however, that this case will require a resolution of this difficult question. That turns on whether we can say that Eichelberg's negligence was last in time as a matter of law. If we can, then the issue would be squarely presented. If we cannot — if a jury could reasonably find that Amtrak's negligence was last in time — then, as we have already indicated, Eichelberg would be entitled to have a jury determine the parties' comparative negligence (although he is by no means assured of ultimate success).

The record is obscure as to the critical question of which party had the true last clear chance to avoid the accident. Whether Eichelberg's negligence was last in time — that is, whether Eichelberg could have avoided injury by exercising reasonable care after the train could no longer have avoided the collision by braking — depends on a number of factors, including: (a) the point in time at which Eichelberg should have realized that (as he claims) the train was not braking; (b) how long it would have taken him, at that point, to reach complete safety by crossing to the other track; (c) whether a reasonable person would have believed that the narrow mezzanine between tracks provided a safe retreat, and if so how long it would have taken Eichelberg to reach the mezzanine; (d) whether it was reasonable for Eichelberg to believe, at the point when he should have realized that the train could no longer help by braking, that he could safely get to the end of the trestle; and (e) whether Eichelberg was still coiling his fishing line as he attempted to leave the trestle, and, if he was, whether it was reasonable under the circumstances for him to do so. As to these factors, the record is insufficiently developed to permit a resolution as a matter of law at this time.

In time, Connecticut courts will undoubtedly shed light on the complex relationship between the rule of comparative negligence and the factors that inhered in the now abandoned doctrine of last clear chance. At the moment, however, we believe that a summary judgment on the current record cannot lie.

III. CONCLUSION

Because we conclude that the district court erred in its application of Connecticut's law of liability to trespassers, and because, on the present record, we cannot say that Amtrak is entitled to judgment as a matter of law under the Connecticut comparative negligence statute, we vacate the district court's entry of summary judgment in favor of Amtrak and remand for further proceedings consistent with this opinion.