Plaintiff-appellant Wilma Williams appeals from a judgment of the United States District Court for the Eastern District of New York (I. Leo Glasser, Judge) in favor of KFC National Management Company ("KFC" and formerly known as "Kentucky Fried Chicken") dismissing her personal injury claim. The District Court concluded that Williams failed to raise a material issue of fact as to KFC's responsibility for a sidewalk condition that allegedly caused her to fall and suffer injuries. After Williams's time to appeal expired, she moved for an extension, which was unopposed. The District Court granted the extension. After the appeal was fully briefed in this Court, KFC moved to dismiss it as untimely, contending that the extension was an abuse of discretion. For the reasons that follow, we conclude that in the absence of an objection to the motion for extension of time, we cannot say that the District Court abused its discretion in granting the extension. Accordingly, KFC's motion to dismiss must be denied. Because we also conclude that Williams raised genuine issues of material fact, we vacate the summary judgment and remand for further proceedings.

The facts considered in the light most favorable to Williams indicate that on September 19, 2000, she was walking down Lafayette Avenue towards Bedford Avenue in Brooklyn, N.Y. The evening was dark, it was raining heavily, and she fell on a section of sidewalk abutting a locked gate that separates the sidewalk from a dumpster storage area located behind a KFC restaurant. According to the assistant manager of the restaurant, about twice a day KFC employees place trash bags filled with refuse such as food containers, food preparation materials, and uneaten food into the dumpster. Overnight, the gate to the dumpster area is left unlocked so that the dumpsters can be removed by refuse carters who drag them through the unlocked gate and across the sidewalk to be emptied into garbage trucks. In the morning, KFC employees drag the dumpsters back over the sidewalk to the storage area.

Williams claims that, after her fall, she discovered grease on her clothes and shoes, and that a police officer who responded to the accident told her she had slipped on grease on the sidewalk. A private investigator retained by Williams returned to the scene two days later, noted that the area was slippery, and took photographs showing discoloration on the stretch of sidewalk where the fall occurred. In addition, the assistant manager testified that on the date in question, the area where the dumpster was located was dirty, and that the garbage bags broke on occasion.

Williams sued KFC in New York state court, and KFC removed to federal court. In her complaint, Williams alleged that twice a day KFC rolled its dumpster across the stretch of sidewalk where she fell, that grease leaked out of the garbage bags and the dumpster, causing slippery conditions on the sidewalk, and that these slippery conditions caused her fall and her injuries. Her alternate theory of liability was that since KFC made "special use" of the sidewalk, under New York law it had actual or constructive knowledge of, and responsibility for, the dangerous condition, even if she had not proved that KFC affirmatively caused it.

After discovery, KFC moved for summary judgment, which the District Court granted. The Court concluded that, although Williams had raised a genuine issue of material fact as to the existence of grease on the sidewalk, she had raised none as to KFC's responsibility for the grease, finding Williams contentions in this regard too speculative. The Court also concluded that Williams had failed to establish that KFC was liable as a consequence of any "special use" of the sidewalk.

Judgment was entered on January 31, 2003, but Williams failed to appeal within the 30 days allowed by the Federal Rules of Appellate Procedure. See Fed. R.App. P. 4(a)(1)(A). Rather, on March 26, she moved for an extension of time to file the appeal. See Fed. R.App. P. 4(a)(5). KFC received the motion on March 27 but failed to oppose it. The District Court granted the motion and Williams filed her notice of appeal on March 31. Months later, on November 19, after appellate briefing had been completed, KFC moved to dismiss the appeal for lack of jurisdiction because of Williams's late filing. We reserved decision. We now deny KFC's motion to dismiss and, reaching the merits, conclude that Williams raised a genuine issue of material fact as to KFC's responsibility for the accident. We therefore vacate the judgment and remand for further proceedings.

Under Rule 4(a), a notice of appeal in a civil case must be filed within 30 days after entry of judgement. Fed. R.App. P. 4(a)(1)(A). The district court may extend that time period if (i) a party moves for the extension no later than 30 days after the time prescribed by Rule 4(a) expires and (ii) the moving party establishes excusable neglect or good cause. Fed. R.App. P. 4(a)(5).

Compliance with Rule 4(a) is "mandatory and jurisdictional." Browder v. Director, Dep't of Corr., 434 U.S. 257, 264, 98 S. Ct. 556, 54 L. Ed. 2d 521 (1978); Silivanch v. Celebrity Cruises, Inc., 333 F.3d 355, 363 (2d Cir.2003). This Court has previously explained that "[t]he power of the federal courts to extend this time limitation is severely circumscribed." Endicott Johnson Corp. v. Liberty Mut. Ins. Co., 116 F.3d 53, 56 (2d Cir.1997). We review for abuse of discretion a district court's decision to grant or deny an extension of time to file a notice of appeal, but if an appeal is filed outside the time limitations provided in Rule 4(a)(5), we lack jurisdiction. Goode v. Winkler, 252 F.3d 242, 245 (2d Cir.2001).

Williams's motion sought to establish "excusable neglect" solely on the ground that "plaintiff's counsel inadvertently closed plaintiff's case after ... the Court had dismissed her case and the plaintiff failed to instruct [her counsel] to file a Notice of Appeal" and that "plaintiff, recently, on March 21, 2003, contacted [her counsel] and indicated that she wanted [counsel] to file a Notice of Appeal herein." Appellant Mot. for Ext. of Time. The motion further contended that Williams had "good cause for the instant appeal" because the evidence demonstrated that "there is a material question of fact concerning the creation of the defect on the defendant's premises that caused plaintiff to fall and sustain injuries." Id. This latter contention was, of course, irrelevant since the appropriate inquiry is not whether the underlying claim has merit, but whether excusable neglect or good cause exists for the failure to file the notice of appeal in a timely manner. As previously noted, none of Williams's excuses were contested and her motion, without opposition, was granted.

The Supreme Court, in Pioneer Investment Services Co. v. Brunswick Associates Ltd. Partnership, 507 U.S. 380, 113 S. Ct. 1489, 123 L. Ed. 2d 74 (1993), adopted a liberal test for assessing what neglect is excusable, emphasizing that "the determination is at bottom an equitable one, taking account of all relevant circumstances," including: (1) the danger of prejudice to the non-moving party, (2) the length of delay and impact on judicial proceedings, (3) the reason for the delay, including whether it was within the reasonable control of the moving party, and (4) whether the moving party acted in good faith. Id. at 395, 113 S. Ct. 1489. To be sure, the District Court could readily have concluded that Williams met three aspects of this four-factor test. In the absence of any objection on KFC's part, there is nothing to suggest prejudice to the defendant due to the three-week delay. Further, the District Court was, of course, competent to conclude that the delay did not affect judicial proceedings. Finally, a finding of good faith on Williams's part was warranted given the District Court's familiarity with the case and the lack of any evidence to the contrary adduced by KFC. We have emphasized, however, that it is the third factor-the reason for the delay-that predominates, and the other three are significant only in close cases. But we have emphasized this factor solely in cases where the non-moving party made an objection to the validity of the explanation given for the delay. Silivanch, 333 F.3d at 366.

Thus, in Silivanch, we held that it was an abuse of discretion for the lower court to have found reliance on an opposing counsel's erroneous statement about a filing deadline to be "excusable neglect." Id. at 370. We expressed concern that "the legal system would groan under the weight of a regimen of uncertainty in which time limitations were not rigorously enforced" and, in that case, where vigorous opposition to the explanations for the neglect was presented, we reaffirmed that the "`excusable neglect standard can never be met by a showing of inability or refusal to read and comprehend the plain language of the federal rules.'" Id. at 368-69 (quoting Weinstock v. Cleary, Gottlieb, Steen & Hamilton, 16 F.3d 501, 503 (2d Cir.1994), and United States v. Hooper, 43 F.3d 26, 29 (2d Cir.1994)) (cases in which forfeiture and failure to object were not at issue). KFC argues that our case law compels the conclusion that the District Court's extension was an abuse of discretion because Williams's only reason for her lateness does not meet the standards of Silivanch.

While on the record before us we cannot say that the reason for delay would fall under Silivanch's definition of excusable neglect, we note a critical difference: in Silivanch the motion for an extension of time — specifically the adequacy of the third Pioneer factor — was vigorously challenged through the filing of opposition papers and participation in oral argument. Silivanch, 333 F.3d at 362. Here, KFC did nothing. It did not respond to Williams's motion, it did not request oral argument, and it said nothing after the District Court granted the extension. KFC's objection came months later, after the notice of appeal was filed and after the appeal had already been briefed to this Court. As set forth below, we hold that the failure to oppose, in circumstances where the other three Pioneer factors could readily be found to have been met, created a context in which the District Court's discretion to grant the motion without further examination was not an abuse of discretion.1

To promote finality and efficiency in appellate proceedings, "[w]e operate in an environment ... in which substantial rights may be, and often are, forfeited if they are not asserted within time limits established by law." Silivanch, 333 F.3d at 367. In Silivanch, that reasoning meant that a party that failed to appeal on time and whose claim of excusable neglect was traversed and, as a matter of law, deemed inadequate with respect to the third Pioneer factor lost the right to appeal. We further held that in such circumstances the meeting of the other Pioneer factors did not suffice to place an extension decision within the district court's discretion. While we adhere to the logic of Silivanch, we find that it leads to the opposite conclusion in the absence of any questioning in the district court of the movant's claim that she has met the third Pioneer factor. We reach this conclusion at least where the other three Pioneer factors can easily be found to run in the movant's favor.

Rule 4(a) requires that requests for extensions be made in the district court. Fed. R.App. P. 4(a). This requirement insures that the court most familiar with the litigation, and in the best position to assess the merits of a request, rule in the first instance. The lack of an objection in the district court to an extension request deprives the district court of the basis for a focused exploration of the sufficiency of the request, and limits the ability of courts of appeals to determine whether granting one was an appropriate exercise of discretion by the district court. See Fed. R.App. P. 4(a)(5)(B) (requiring notice to the other party if the motion for extension is filed after the prescribed time for appeal, allowing an opportunity for opposition). Since this determination would be made with no answering or reply papers and without the benefit of the views of the district court developed in a contested proceeding, it would be, in all likelihood, a less informed decision, one that unnecessarily enhances the risk of an incorrect result. But the alternative, to send the matter back to the district court for further findings, would give rise to precisely the sort of delay and uncertainty (as a result of appellee's failure to act in a timely fashion) that Rule 4(a) was designed to avoid.

More fundamentally, allowing appellees belatedly to question the District Court's extension as an abuse of discretion at any time (since the objection goes to jurisdiction) invites manipulation. It allows the appellee to "wait and see," and raise the jurisdictional issue — which is one the appellate court can often discern on its own only with difficulty — only if the oral argument or perhaps even the appellate decision has "gone against" them. Such an incentive is to be avoided if at all possible.

On the facts before us, avoiding such an undesirable result is readily possible. The standard for reviewing the grant of a motion to extend is abuse of discretion. Three of the four Pioneer factors, on their face, support the court's ruling in favor of that extension. The "reason for delay" factor, concededly the most important, does not support an extension, but its adequacy was not challenged before the District Court. Where the validity of the this factor goes unchallenged, and the other factors favor extension, we cannot say that the District Court abused its discretion by granting the extension or in deciding not to, sua sponte, examine the facts in more depth.

We review a summary judgment de novo. Hellstrom v. United States Dep't of Veterans Affairs, 201 F.3d 94, 97 (2d Cir.2000). The District Court granted KFC summary judgment on the theory that while Williams raised a genuine issue of material fact as to the existence of a hazardous condition on the sidewalk, she failed to raise one as to KFC's responsibility for the condition. On appeal, Williams contends that she raised these issues of fact in two ways: (1) with respect to whether KFC created, and consequently was responsible for, the hazardous conditions that caused her injuries, and (2) whether, under New York law, KFC made "special use" of the sidewalk and therefore could be charged with constructive knowledge of, and responsibility for, the hazardous condition.

With respect to "special use," it is unclear that this argument was raised below. The District Court noted that "[p]laintiffs do not allege or argue that KFC made any special use of the sidewalk" but, in any case, "there is no evidence that KFC owns, occupies, controls, or makes special use of the sidewalk." On appeal, Williams contends that the special use argument was raised below when she averred that the condition creating the cause of the fall was the movement of the dumpster across the sidewalk, and that KFC had actual or constructive knowledge of this condition.

While it is true that if the question of "special use" had not been raised before the District Court, it would have been waived here, United States v. Keppler, 2 F.3d 21, 23 (2d Cir.1993), we need not definitively resolve this question since it is as clear to us as it was to the court below that the "special use" doctrine has no place in this case. Accordingly, we can proceed on the assumption that the District Court granted summary judgment with respect to the special use question as well as to the affirmatively-caused-defect theory. Fama v. Comm'r of Correctional Servs., 235 F.3d 804, 816 n. 11 (2d Cir.2000) (noting that courts may assume hypothetical jurisdiction where a jurisdictional requirement is not constitutional).

Under New York law, it is well established that liability for injuries resulting from negligently maintained public sidewalks lies, if at all, with the municipality, not the adjacent landowner. Hausser v. Giunta, 88 N.Y.2d 449, 452-53, 646 N.Y.S.2d 490, 669 N.E.2d 470 (1996). However, exceptions exist, and an abutting owner will be presumed to have constructive knowledge of dangerous conditions and may be liable for resulting injuries: (1) where the sidewalk was constructed in a special manner for his benefit, in other words, where he makes "special use" of the sidewalk, (2) where he affirmatively caused the defect, (3) where he negligently constructed or repaired the sidewalk, and (4) where a local ordinance specifically requires the landowner to maintain the sidewalk and imposes liability for failure to do so. Hausser, 88 N.Y.2d at 453, 646 N.Y.S.2d 490, 669 N.E.2d 470. On appeal, Williams contends that KFC made "special use" of the sidewalk by dragging the dumpster over the sidewalk and that this use was sufficient to impose a duty of care on KFC that it breached.

We find this argument unconvincing. In New York, the "special use exception is reserved for situations where a landowner whose property abuts a public street or sidewalk derives a special benefit from that property unrelated to the public use, and is therefore required to maintain a portion of that property." Poirier v. City of Schenectady, 85 N.Y.2d 310, 315, 624 N.Y.S.2d 555, 648 N.E.2d 1318 (1995). When considering the doctrine, New York courts have consistently required that "[b]efore liability can be imposed, the sidewalk must be constructed in a special manner" for the benefit of the abutting landowner. Kiernan v. Thompson, 137 A.D.2d 957, 958, 525 N.Y.S.2d 380 (N.Y.App. Div., 3d Dep't 1988) (gathering cases). Cases applying the doctrine have typically involved the installation of some object in the sidewalk or a variance in the construction of the sidewalk intended specifically to benefit the adjacent owner. See, e.g., Granville v. City of New York, 211 A.D.2d 195, 627 N.Y.S.2d 4 (N.Y.App. Div., 1st Dep't 1995) (involving a concrete step mounted on the sidewalk beneath the elevated doorway of a restaurant); Santorelli v. City of New York, 77 A.D.2d 825, 430 N.Y.S.2d 618 (N.Y.App. Div., 1st Dep't 1980) (involving a heating oil filler cap in the sidewalk); Nickelsburg v. City of New York, 263 A.D. 625, 34 N.Y.S.2d 1 (N.Y.App. Div., 1st Dep't 1942) (involving iron bars that were embedded in the sidewalk to facilitate removal of refuse). Here, Williams does not allege that any special features on the sidewalk were constructed for KFC's benefit.

Significantly, New York courts have repeatedly refused to find a special use even when abutting landowners have made considerably more intrusive use of the sidewalk than did KFC. See, e.g., Thomas v. Triangle Realty Co., 255 A.D.2d 153, 679 N.Y.S.2d 394 (N.Y.App. Div., 1st Dep't 1998) (finding that maintenance of an ATM machine adjacent to sidewalk was not a special use); Tortora v. Pearl Foods, Inc., 200 A.D.2d 471, 606 N.Y.S.2d 235 (N.Y.App. Div. 1st Dep't 1994) (finding that a business whose customers formed a line on the sidewalk was not making a special use of the public property). In contrast, KFC's use of the sidewalk to move a dumpster was not special; it was an entirely routine use, indistinguishable from the myriad ways in which public sidewalks are used every day. Consequently, the special use doctrine does not apply.

Alternatively, Williams claims that KFC is liable because it affirmatively caused the dangerous condition. The District Court found — crediting Williams's testimony as to the grease she found on her clothes, the statements of the responding police officer, and the testimony of the private investigator — that Williams had presented sufficient evidence to raise a genuine issue of material fact as to the existence of the hazard. The District Court went on, however, to conclude that finding KFC responsible for the condition required an accumulation of speculative inferences: (1) that the garbage bags placed inside the dumpster had leaked, (2) that the dumpster itself had leaked, (3) that the dumpster had leaked while it was being moved across the sidewalk, and (4) that leaks during this movement left a greasy residue on the sidewalk. Finding these inferences too attenuated, the District Court concluded that Williams had failed to raise a genuine issue of material fact as to KFC's responsibility for the hazardous condition of the sidewalk. We disagree.

Under New York law, Williams was not required to adduce the most reasonable explanation for the accident, nor was she required to eliminate all other possible causes for her fall. To avoid summary judgment, she was obligated simply to establish a reasonable probability that the accident was caused by KFC's negligence. In Ingersoll v. Liberty Bank of Buffalo, 278 N.Y. 1, 14 N.E.2d 828 (1938), the New York Court of Appeals considered a negligence claim arising from a defectively repaired stairway. The Appellate Division had dismissed the complaint, finding that the plaintiff had proven a defect but had failed to show a causal connection between the defect and the injury because the defendant had been able to proffer an alternative explanation for the injury. Id. at 5-6, 14 N.E.2d 828. The Court of Appeals reversed:

Where the facts proven show that there are several possible causes of an injury, for one or more of which the defendant was not responsible, and it is just as reasonable and probable that the injury was the result of one cause as the other, plaintiff cannot have a recovery, since he has failed to prove that the negligence of the defendant caused the injury. This does not mean that the plaintiff must eliminate every other possible cause. The plaintiff was not required to offer evidence which positively excluded every other possible cause of the accident. The existence of remote possibilities that factors other than the negligence of the defendant may have caused the accident, does not require a holding that plaintiff has failed to make out a prima facie case. It is enough that he shows facts and conditions from which the negligence of the defendant and the causation of the accident by that negligence may be reasonably inferred.

Id. at 7, 14 N.E.2d 828 (internal citations and quotation marks omitted). The Court found that "the natural and reasonable inference is that the plaintiff was descending the stairway with the box, when the defective tread broke under his foot and caused him to fall," id. at 8, 14 N.E.2d 828, and that the defendant's explanation, that the plaintiff suffered a heart attack and fell, was but a "remote probability." Id. at 6, 14 N.E.2d 828. Even though there was evidence that the plaintiff suffered from a heart condition that could have caused the accident, the Court found this possibility insufficient to eliminate, at least at the summary judgment phase, the competing inference that the defective repairs caused the accident. Id.

The New York Court of Appeals has applied this reasoning with a fair degree of consistency when evaluating the adequacy of circumstantial evidence in negligence cases. For example, in considering the case of a hospital patient who fell out of her bed because its guardrails had been lowered, the Court stated:

To establish a prima facie case of negligence based wholly on circumstantial evidence, "[i]t is enough that [plaintiff] shows facts and conditions from which the negligence of the defendant and the causation of the accident by that negligence may be reasonably inferred." The law does not require that plaintiff's proof "positively exclude every other possible cause" of the accident but defendant's negligence. Rather, her proof must render those other causes sufficiently "remote" or "technical" to enable the jury to reach its verdict based not upon speculation, but upon the logical inferences to be drawn from the evidence.

Although plaintiff may in her attempt to meet that burden include proof tending to negate the significance of other possible causes, we have on numerous occasions upheld or reinstated a jury's verdict where the logic of common experience itself, as applied to the circumstances shown by the evidence, led to the conclusion that defendant's negligence was the cause of plaintiff's injury.

Schneider v. Kings Highway Hosp. Ctr., Inc., 67 N.Y.2d 743, 744-45, 500 N.Y.S.2d 95, 490 N.E.2d 1221 (1986) (internal citations omitted). The Court found that the plaintiff had established a prima facie case through evidence that the hospital had a policy requiring guardrails to be raised for all patients over 70 and that it was more likely for a hospital worker than the patient herself to have lowered the rails. Even though the defendant offered evidence that the patient had previously lowered the rails, the Court concluded that the inconvenient location of the mechanism that adjusted the rails and plaintiff's frail condition made it more likely that a hospital employee had lowered them. Id. at 745, 500 N.Y.S.2d 95, 490 N.E.2d 1221. Significantly, according to the Court, the plaintiff was "not required to prove the exact nature of defendant's negligence." Id. Similarly, in Gayle v. City of New York, 92 N.Y.2d 936, 680 N.Y.S.2d 900, 703 N.E.2d 758 (1998), the Court of Appeals considered the claim, based on circumstantial evidence, that a large puddle on a roadway resulting from a negligently maintained drainage system was the proximate cause of an accident. The Court concluded that the Appellate Division had "erred in determining that plaintiffs were required to rule out all plausible variables and factors that could have caused or contributed to the accident." Id. at 937, 680 N.Y.S.2d 900, 703 N.E.2d 758. Following Schneider, the Court emphasized that the plaintiff did not have to exclude all other possible causes of the accident, but only had to offer proof to render those other causes "remote" or "technical." Id. "A plaintiff need only prove that it was `more likely' or `more reasonable' that the alleged injury was caused by the defendant's negligence than by some other agency." Id. (internal citation omitted). Lower courts in New York have also followed this standard. See, e.g., N.Y. Tel. Co. v. Harrison & Burrowes Bridge Contractors, Inc., 3 A.D.3d 606, 771 N.Y.S.2d 187 (N.Y.App. Div., 3d Dep't 2004) (finding evidence that defendant had dropped 5,000 pound steel beams near plaintiff's cables sufficient to withstand summary judgment and holding that plaintiff was not required to establish defendant's liability to an "unassailable certainty").

Applying these principles, we conclude that, for summary judgment purposes, Williams presented sufficient circumstantial evidence to render the probability that someone other than KFC was responsible for the grease sufficiently "remote" or "technical" in comparison to the explanation she offered. As the District Court realized, the grease on Williams' clothing, the statement by the responding police officer as to the presence of grease, and the statements of the private investigator as to greasy spots, all point to the existence of grease on the sidewalk. Additionally, significant testimony from the assistant manager of the restaurant establishing that KFC's trash bags had leaked on occasion, that the bags contained greasy food residue, and that the dumpster area behind the restaurant was dirty on the date of Williams' accident supports the inference that KFC was responsible for the greasy conditions. In light of these facts, we conclude that Williams carried her burden of establishing that a reasonable factfinder could conclude that KFC was responsible for the grease on the sidewalk and that summary judgment was not warranted on the basis of the facts before the District Court.

We deny KFC's motion to dismiss this appeal for lack of jurisdiction. We vacate the judgment of the District Court and remand for further proceedings consistent with this opinion.

CALABRESI, Circuit Judge, concurring.

This case presents a simple fact pattern — a slip-and-fall on a sidewalk outside a restaurant — but implicates complex, and longstanding, issues concerning circumstantial evidence. With that in mind, while I readily concur in the opinion of the Court, I write separately to provide some analysis on the key substantive question underlying this appeal: What does the law of torts tell us about the amount and type of circumstantial evidence that plaintiffs must adduce, in different circumstances, in order to reach a jury?

A review of the doctrine reveals three (closely related) areas in which this sort of determination has proven difficult:

1. There is sufficient evidence that defendant caused plaintiff's injury. Whether the defendant was negligent in doing so, however, is uncertain. This is the situation in which the notion of res ipsa loquitur was first articulated in the 1863 case of Byrne v. Boadle, 159 Eng. Rep. 299 (1863).2

2. There is sufficient evidence that plaintiff was injured on account of negligence. But it is an open question whether that negligence was the defendant's or was, instead, someone else's. See, e.g., Ybarra v. Spangard, 25 Cal. 2d 486, 154 P.2d 687 (1944).

3. The evidence that a particular defendant was negligent is clearly sufficient, as is the evidence that the defendant's acts or activities were implicated in the plaintiff's injury. But the question of whether it was defendant's negligence — rather than just some non-negligent act of the defendant's, or some actions of another party, including the plaintiff — that caused the injury is up for grabs.3

In recent years, all three of these areas have generally been framed in terms of the sufficiency of the circumstantial evidence; that is, if the evidence of A is sufficient, can a jury find, by common experience, that B is more probable than not? But in fact, to understand these cases, one must examine three separate considerations.

A. Consideration One: Strength of Circumstantial Evidence

The first is the straight circumstantial evidence inquiry-how strong is the evidence and how decisively does it tilt us toward accepting the plaintiff's allegations? In res ipsa cases, for example, courts look for "a basis of past experience which reasonably permits the conclusion that such events do not ordinarily occur unless someone has been negligent." Restatement (Second) of Torts § 328D cmt. c (1965). See also W. Keeton, D. Dobbs, R. Keeton & D. Owen, Prosser and Keeton on Law of Torts § 39, at 243 (5th ed.1984) (hereinafter "Prosser and Keeton") (indicating that the key to res ipsa is a reasonable inference, "based upon the evidence given, together with a sufficient background of human experience to justify the conclusion").4 Thus, the "fall of an elevator" or the "explosion of boilers" provides a stronger case for finding negligence on the basis of circumstantial evidence than does "[t]he fact ... that a man falls down stairs." Restatement (Second) of Torts § 328D cmt. c.5

B. Consideration Two: Relative Knowledge of the Parties

The second is an inquiry into which of the parties is in a better position either to reveal or to seek out explanatory evidence. (Such evidence can be direct or circumstantial, but in most cases the focus will be on teasing out direct evidence). This consideration is often cast in terms of which side has more knowledge and therefore which side should bear the incentive to come forward with the evidence. See Griffen v. Manice, 166 N.Y. 188, 59 N.E. 925, 925 (1901) (the doctrine of res ipsa is partly premised on the fact that the defendant has the ability to "produce evidence of the actual cause that produced the accident, which the plaintiff is unable to present."). In this regard, some states have added a requirement to the traditional res ipsa doctrine that the relevant evidence be more accessible to defendants. See, e.g., Wilson v. Stilwill, 411 Mich. 587, 309 N.W.2d 898, 905 (1981) (listing this element as one requirement); Bass v. Nooney Co., 646 S.W.2d 765, 768 (Mo.1983) (same); Goedert v. Newcastle Equip. Co., 802 P.2d 157, 160 (Wyo.1990) (same). Other commentators and courts, however, have understandingly been reluctant to make this aspect a requirement additional to that of sufficient circumstantial evidence of negligence. See 1 Stuart M. Speiser, The Negligence Case: Res Ipsa Loquitur § 2:27 n. 15 (1972) (collecting cases); Prosser and Keeton § 39, at 254-55; Restatement (Second) of Torts § 328D cmt. k (1965).6 Although it may be that this consideration is seldom controlling, see Prosser and Keeton § 39, at 254, it does, as we will see, still have important "persuasive effect" in certain circumstances, id. at 255.

C. Consideration Three: Symmetry of Error Costs

The third is an inquiry into how strongly we feel about making an error in one direction as against the other. That is to say, were we erroneously to find that the evidence was sufficient to reach a jury, would that be as, less, or more harmful than if we were erroneously to find that the evidence was in sufficient? In res ipsa cases in particular, this is sometimes improperly described in terms of how close or how far we are from strict liability. But that is just one manifestation of the question. In Judson, for example, the court was concerned with the prospect of erroneously leaving the plaintiffs' injuries uncompensated. See 40 P. at 1021. In traditional medical malpractice cases, by contrast, there was a concern that even when the circumstantial evidence is relatively strong, holding a doctor liable could prove pretty tough medicine to swallow. See, e.g., Semerjian v. Stetson, 284 Mass. 510, 187 N.E. 829, 831 (1933) (holding that "res ipsa loquitur is not applicable" where a physician placed a liquid substance on the plaintiff's eye, and the eye immediately thereafter became inflamed and pained, and the plaintiff eventually lost all sight in that eye, because "it could not be said that" the plaintiff's condition would not have existed except for negligence). Ultimately, an assessment of the adequacy of the circumstantial evidence was driven in part by a court's intuition about which side should bear the costs of a mistaken assessment as to the strength of that evidence.

With these considerations in mind, I return to the three broad types of cases — which I identified at the outset — where the circumstantial evidence problem has consistently troubled reviewing courts. In all three areas, the trend has been in the direction of easing the burden on plaintiffs, and courts now more readily permit such cases to get to a jury. But the trend has not developed uniformly in the three areas.

The move toward jury determinations happened first and most easily, in the Nineteenth Century, as to the question of whether there was negligence on the part of a defendant who caused plaintiff's injury. While this may seem surprising, it is, in fact, historically understandable because the concept of a defendant's liability based on fault is much less rooted in the common law than the notion that a defendant's liability presupposes some harm actually caused by the defendant. As an anthropological matter, fault — as a requirement for liability — was a relatively new notion in the nineteenth century.7 By contrast, causation has been "a well-recognized and essential element of the plaintiff's case in chief" since at least the early 17th century, see Zuchowicz, 140 F.3d at 384 n. 2, and probably much, much before.8

It is certainly true, however, that the traditional res ipsa formulation called for defendants to be in control. And one might suppose that the "exclusive control" test attempted to mitigate errors in assessing the sufficiency of the circumstantial evidence. Lack of control increases the chances that the accident occurred in an unusual way, rather than through ordinary, and hence circumstantially expected, negligence. More important, however, control implicates causation, and assures that it was the defendant's circumstantially inferred negligence that brought about the harm.

But if either of these were behind the control requirement, the necessity of exclusive control would be overkill. For in torts, plaintiffs need only to prove both that the defendant was negligent, and that negligence caused the harm complained of, more probably than not. Thus, the certainty of causation that exclusive control seems to entail is not generally needed. I therefore believe that the traditional res ipsa approach to "control" rested on the intuition that if a defendant had exclusive control, the defendant was likely to have superior knowledge, and hence even in the absence of strong circumstantial evidence, could explain the accident. In such circumstances, it is easy to explain courts' inclination to allow juries to determine whether defendants were negligent, based on the second consideration — knowledge and information — rather than on the first or third. Consider, for example, Kanter v. St. Louis, Springfield & Peoria R.R., 218 Ill.App. 565 (1920), in which the plaintiff secured a judgment at trial against the defendant railroad company for damages arising from a train derailment that cost the plaintiff's husband his life. The trial court originally used traditional res ipsa notions to push the defendant, who was in sole control, to bring forth evidence as to what happened in the, at the time, unexplained accident. The defendant thereupon demonstrated that "the derailment and death of deceased was the result of acts of vandalism committed by persons unknown" and not the fault of the railroad. Id. But plaintiff ultimately won because on the facts shown by the defendant, a jury could conclude that a careful defendant would have seen the defect which caused the derailment in time to avoid harm. Id.

Control, as a surrogate for knowledge, had done its job, even in the absence of strong circumstantial evidence. For who could truly say in 1921, from common experience, that railroad derailments occur more probably than not, due to the railroad's fault? In light of the role "control" plays in cases like Kanter, it should not be surprising that, as the focus on relative knowledge has become more self-conscious, see supra nn. 4-5, and as the requirements as to causation have eased, see infra Part III.C, the necessity for actual exclusive control has also diminished.9 See majority opinion at Part II.B.1.

The next area to move in the direction of jury determination was the question of who was negligent when the evidence was sufficient to establish that an injury had occurred on account of someone's negligence. One might think that this question posed the most difficulties because, after all, in these cases, the defendant may well have played no role whatsoever in causing the injury. And yet a rule emerged at mid-20th-century whereby plaintiffs could get to a jury if they could show, more probably than not, that their injury was caused by negligence, even though they possessed no direct evidence as to which of several possible parties acted negligently. A classic example was the situation in which defendant's car, parked on the side of the hill, rolled down the hill and injured the plaintiff hours after the defendant had parked the car. The plaintiff could not directly demonstrate that the defendant was negligent — perhaps it was a "tampering stranger," see Prosser and Keeton § 39, at 249 — yet courts generally permitted these cases to reach a jury. Cf. Roberts v. Ray, 45 Tenn.App. 280, 322 S.W.2d 435, 437-38 (1959).

Multiple defendants were a similar, if more complicated, affair. Only in certain circumstances, generally when the group of defendants were in control, or had greater knowledge than the plaintiff, have courts proven willing to allow a jury to determine which among several defendants was the negligent party. In the famous case of Ybarra v. Spangard, 25 Cal. 2d 486, 154 P.2d 687 (1944), for example, the plaintiff allegedly received certain injuries during surgery that seemed clearly to be the result of negligence. Because he was unconscious he could not identify the responsible party. Id. at 689. The court held that "where a plaintiff receives unusual injuries while unconscious and in the course of medical treatment, all those defendants who had any control over his body or the instrumentalities which might have caused the injuries may properly be called upon to meet the inference of negligence by giving an explanation of their conduct." Id. at 691.10

Several jurisdictions have adopted the Ybarra approach, in the medical context and occasionally in others. Thus, in Stone v. Courtyard Mgmt. Corp., 353 F.3d 155, 160 (2d Cir.2003), we concluded: "It is impossible for the plaintiff to say which of the[ ] two defendants was at fault in this case (if either), but that is precisely why New York law allows plaintiff to employ the res ipsa loquitor inference against each defendant and let them explain what happened." Id. at 160. See also Schroeder v. City & County Sav. Bank of Albany, 293 N.Y. 370, 57 N.E.2d 57, 59 (1944); but see Barrett v. Emanuel Hosp., 64 Or.App. 635, 669 P.2d 835, 838 (1983) (noting that Ybarra's goals could "be achieved in various direct ways which may warrant societal consideration ... [but] we do not think the objective should be pursued by stretching a permissible inference beyond the point where there are underlying facts other than the result from which it can reasonably be drawn").

Ybarra, whatever its merits, provides important insight into the considerations at play in determining who is negligent once negligence has been established. That the plaintiff in Ybarra was injured on account of negligence, or that a jury could readily have so found, was, to the Ybarra court, a very significant fact. As a result, the court focused on the third consideration — that an "insufficiency" error in favor of a defendant might work a greater injustice — result in a greater cost misallocation — than a "sufficiency" error in favor of the plaintiff. Of course, the second consideration was not absent from the analysis; the court relied in part on the fact that certain types of evidence — such as that going to who did wrong — at issue in Ybarra are "practically accessible to [defendants] but inaccessible to the injured person." 154 P.2d at 689 (internal quotes and citations omitted).11 But, in the end, the intuition driving the California rule seemed to be that in its absence a patient who suffered "permanent injuries of a serious character, obviously the result of some one's negligence, would be entirely unable to recover unless the doctors and nurses in attendance voluntarily chose to disclose the identity of the negligent person and the facts establishing liability." Id. That, the court concluded, would be a "gross injustice." Id. To put it another way, an error leaving the loss on the plaintiff would be more serious than one that placed the loss incorrectly on some of the parties in the group responsible for the operation. Id.12

This movement more readily to send to a jury cases in which the plaintiff lacks direct evidence of the particular defendant's negligence is, of course, a limited one.13 Nevertheless, Ybarra's bottom-line instinct about an "injustice" to the negligently injured plaintiff, readily shared in other jurisdictions, often shapes the amount of circumstantial evidence of individual wrongdoing that courts demand in this context.

The area in which there was the most persistent reluctance to send cases to the jury was the third: where the evidence that the defendant was negligent was sufficient, but where the evidence that the plaintiff was injured on account of that negligence was seemingly weak.14 That is, situations in which it often was not clear whether (1) negligent behavior (2) of the defendant was a cause of the injury. Only recently has a consensus developed that such cases should go to a jury upon a relatively light showing by the plaintiff of but for causation. Interestingly, courts in this context have focused not so much on the question of knowledge or access to information, or of "bias" for or against liability, as on the first consideration — whether the circumstantial evidence of but for cause is strong enough. Specifically, did defendant's negligent actions significantly increase the chances that the injury at issue would occur?

The first significant movement in the direction of letting this type of case go to a jury came as early as 1920, when Judge Cardozo articulated a highly significant causation test in Martin v. Herzog, 228 N.Y. 164, 126 N.E. 814 (1920).15 Martin involved a nighttime collision between two vehicles, one of which was operating without the headlights required by law. Judge Cardozo pointed out that lights had been mandated precisely to avoid the risk of such accidents. As a consequence, it could be inferred that the negligent absence of lights was in fact a cause of the accident. The burden therefore shifted to the defendant to demonstrate that some other element had been responsible for the particular crash in question. We have characterized Judge Cardozo's general formulation as follows: If (1) a negligent act "was deemed wrongful because that act increased the chances that a particular type of accident would occur," and (2) "a mishap of that very sort did happen, this was enough to support a finding by the trier of fact that the negligent behavior caused the harm." Zuchowicz, 140 F.3d at 390 (emphasis omitted).

But Martin did not settle the debate. Some New York courts, as well as some in other jurisdictions, proved reluctant to accept such circumstantial evidence as sufficient in these sorts of cases. Such judges continued to fret over the logical fallacy of post hoc ergo propter hoc ("after this, therefore, because of this"), and persisted in demanding direct evidence of causation. See, e.g., Wolf v. Kaufmann, 227 A.D. 281, 237 N.Y.S. 550, 551 (N.Y.App.Div.1929) (denying recovery for death of plaintiff's decedent, who was found unconscious at foot of stairway which, in violation of a statute, had been left unlighted by the defendant, because the plaintiff had offered no proof of "any causal connection between the accident and the absence of light").

Over time, however, the Martin view came to hold sway. In Clark v. Gibbons, 66 Cal. 2d 399, 58 Cal. Rptr. 125, 426 P.2d 525 (1967), for example, the California court referred to the "the low incidence of [such] accidents when due care is used ... combined with proof of specific acts of negligence of a type which could have caused the occurrence complained of" in concluding that a showing on both of these accounts sufficed to permit the case to go to the jury. Id. at 534.16 Today, the broad acceptance of this approach is reflected in and explained by Prosser and Keeton:

And whether the defendant's negligence consists of the violation of some statutory safety regulation, or the breach of a plain common law duty of care, the court can scarcely overlook the fact that the injury which has in fact occurred is precisely the sort of thing that proper care on the part of the defendant would be intended to prevent, and accordingly allow a certain liberality to the jury in drawing its conclusion.

Prosser and Keeton § 41, at 270. See also Guido Calabresi, Concerning Cause and the Law of Torts: An Essay for Harry Kalven, Jr., 43 U. Chi. L.Rev. 69, 71-73 (1975) (examining the three requirements for establishing that the violation of a safety regulation — or for that matter, the contravention of any duty — was a legal "cause" of a plaintiff's injury).

To date, in the causation context, the critical inquiry has seemed to be less focused on who has the most knowledge, or whom the court is inclined to protect from errors. Rather, it has been whether the circumstantial evidence is strong enough. Most of the cases in this area that have relaxed the plaintiff's proof requirements have involved fact patterns where it is reasonable to infer from the available evidence that the defendant's negligent acts were in fact a but for cause of the injury at issue. They have, for the most part, been cases in which the harm that occurred is precisely the harm that the community would expect to occur given a certain negligent action. See, e.g., Liriano v. Hobart Corp., 170 F.3d 264, 271 (2d Cir.1999) ("When a defendant's negligent act is deemed wrongful precisely because it has a strong propensity to cause the type of injury that ensued, that very causal tendency is evidence enough to establish a prima facie case of cause-in-fact."). On the other hand, where that inference cannot so easily be drawn — that is, where that expectation is less clear and the circumstantial evidence therefore weaker — many courts continue to remain reluctant to allow the causation question to reach a jury. That said, the other two strands or factors — relative knowledge, and asymmetry in the significance of error — are not absent in this area either. They seem certainly to have played a role in cases like Clark and Zuchowicz, and may well come in time to be as significant in this context as they are in the other two.17 All three factors are, after all, as relevant in this context as they are in the areas in which they have already come to play an important role.

Having examined the treatment of circumstantial evidence in various areas of tort law, we are left with the question: In which category does the case before us belong? What makes the instant case difficult is that the evidence that the plaintiff was injured on account of negligence, while enough to get to a jury, is just barely so. As the district judge — an eminent and distinguished tort scholar — put it, "although the evidentiary record is scant, it is colorable that a genuine issue of material fact does exist as to whether" there was "grease on the sidewalk where she fell at the time of the accident." [A 211]. Once that has been established, the case seems to fit into the second category — whose negligence was it?18

The case thus bears resemblance, as the majority opinion notes, to Schneider v. Kings Highway Hosp. Center, 67 N.Y.2d 743, 500 N.Y.S.2d 95, 490 N.E.2d 1221 (1986), in which plaintiff's decedent, an elderly hospital patient, fell out of her bed because the side rails had been lowered. There was no question that whoever lowered the rails had committed a negligent act, and that the negligent act was a cause of the resulting injury. The critical inquiry there, as here, centered on whether the negligent party was the defendant or someone else. On the court's view, the lack of direct evidence of who raised or lowered the rails was not fatal to the plaintiff's case: "Although plaintiff may in her attempt to meet that burden include proof tending to negate the significance of other possible causes, we have on numerous occasions upheld or reinstated a jury's verdict where the logic of common experience itself, as applied to the circumstances shown by the evidence, led to the conclusion that defendant's negligence was the cause of plaintiff's injury." See id. at 1221 (internal citation omitted).

Here, the logic of common experience tends to support allowing this case to go to a jury. It is clear from the record that KFC is the most likely spiller of grease.19 Despite the district court's emphasis on the number of necessary inferences, the basic question a fact-finder must answer is a simple one: given that this patch of sidewalk is traversed by a dumpster with garbage bags full of greasy refuse, could a reasonable factfinder conclude that KFC is the source of the grease on the sidewalk? Like the rest of the panel, I believe that under New York law, the answer is yes. KFC rolled its dumpsters from a "dirty" area behind the restaurant, across the relevant patch of sidewalk, and to the street. Those dumpsters contained bags holding the remains of greasy food. Those bags occasionally "burst." What is more, there is no other clearly plausible candidate for this negligent act. Thus, the other potential explanations are "sufficiently remote [and] technical to enable the jury to reach its verdict based" upon something other than speculation. Id. (internal quotation marks omitted); see also Johnson v. New York City Transit Auth., 129 A.D.2d 424, 513 N.Y.S.2d 687 (N.Y.App.Div.1987) (citing Schneider and relying on circumstantial evidence to reverse the dismissal of a case where plaintiff had been injured by a flying object that appeared to come from defendant's subway train passing by overhead).

All in all, given the strength of the circumstantial evidence in this case, the relative capacity of the parties to explain how the offending grease got on the sidewalk, and the absence of any reason to prefer erring in favor of KFC rather than the plaintiff, I am convinced that the result we reach today is not only mandated by New York law but is also consistent with the modern doctrinal trends at the complicated intersection of circumstantial evidence and tort law.