POSNER, Circuit Judge.
This civil rights suit under 42 U.S.C. § 1983 charges members of the Chicago Police Department with having entered and [1563]*1563searched the plaintiffs’ home, and seized the plaintiffs, in violation of the Fourth Amendment (held applicable to state action by virtue of the Fourteenth Amendment), which, in its first clause, guarantees “the right of the people to be secure in their persons [and] houses, ... against unreasonable searches and seizures.” The jury brought in a verdict for the defendants, and the plaintiffs appealed. A panel of this court (with Judge Pell in partial dissent) held that the district judge should have directed a verdict for the plaintiffs. 739 F.2d 1186 (7th Cir.1984). The full court ordered rehearing en banc to consider the division of functions between judge and jury in a damage suit charging violations of the Fourth Amendment. Differently constituted majorities of the court have now concluded that although the trial judge was right not to grant the plaintiffs’ motion for directed verdict (except with respect to the 42-hour detention of David Llaguno), he committed errors that entitle the plaintiffs to a new trial.
On a night in Chicago in 1980, two young Hispanic men committed two robberies, killed four people and wounded three others (including a policeman), and abducted a young girl. When the getaway car crashed, the police were able to shoot and capture one of the killers (Garcia, who has since been sentenced to death) and recover the girl unharmed, but the other killer escaped on foot. A check of the license-plate number showed that the car was registered to Vilma Llaguno at an address two miles from the crash site and that it had not been reported stolen. The crash occurred at North and Oakley; Vilma Llaguno’s address was on Wabansia, near North Avenue but farther west than Oakley. One of the robberies had taken place between the crash site and the Llaguno residence.
Several policemen, led by Sergeant Mingey, drove to their headquarters, picked up a shotgun and a sledgehammer there, and then drove to the Llaguno home, believing that the killer who had fled from the car when it crashed may have been living at Vilma Llaguno’s address, and that fleeing felons often go home. (Mingey and several other policemen in the entry party are one group of defendants; the other consists of policemen involved in the protracted detention of David Llaguno, of which more shortly.) Upon arrival Mingey banged on the front door and ordered the woman who came to the door, Gloria Llaguno, to open it. She did so, and the police rushed in with drawn guns, searched the house, rounded up the occupants (the plaintiffs in this action), and herded them into the living room. Those seized included Gloria and her husband, several of their children (including David Llaguno), and several grandchildren — a total of 10 people. (Vilma Llaguno, who is Gloria Llaguno’s daughter-in-law, was not at home.) In response to questions from the police, David revealed that it was his car that had crashed, and said he had loaned it to a friend. When the police asked him who the friend was, he gave Garcia’s name, according to David’s testimony; according to the police, he refused to answer. They arrested him. Some of the plaintiffs testified at trial that the police threatened to shoot them, which the police denied; that the police had later come back to the house to speak to David; and that on these occasions they had entered the house without anyone’s consent, which they also denied — while acknowledging having held David in custody for 42 hours after his arrest, during which time they neither charged him with a crime nor brought him before a magistrate.
While the police were at the Llaguno residence, the killer who had fled from the crash at North and Oakley was shot and killed by other policemen. He turned out to be Roger Llaguno, a son of Gloria and brother of David but not a resident of the house that the police had entered. No charges were ever lodged against any of the occupants, including David.
The plaintiffs argue that even if the police had probable cause to search the house and detain its occupants (an issue we shall come back to), they still violated the Fourth Amendment as a matter of law by [1564]*1564failing to get a search warrant. Except in an emergency (“exigent circumstances”), police may not, with neither a warrant nor the homeowner’s permission, search a home even though they have probable cause to believe a search would be fruitful. See, e.g., Welsh v. Wisconsin, — U.S. -, 104 S.Ct. 2091, 2097, 80 L.Ed.2d 732 (1984). But if the police in this case had probable cause to believe that the killer was in the Llaguno house, they were excused from getting a warrant, which could have imposed a delay of several hours. The situation was an emergency in about as vivid a sense as can be imagined. A man had (with his partner) just shot seven people. There was no reason to think he had finished shooting; there was every reason to think he would put up a violent resistance. If the police delayed for a warrant, the killer might barricade the house, take hostages, or flee and kill again before they could catch up with him.
True, other cases where an emergency has been held to justify a search without a warrant have involved a clearer showing of probable cause for the search than this case. See, e.g., Warden v. Hayden, 387 U.S. 294, 297-99, 87 S.Ct. 1642, 1645-1646, 18 L.Ed.2d 782 (1967); Dorman v. United States, 435 F.2d 385, 393 (D.C. Cir.1970) (en banc). Yet even Dorman, which lists six factors to be considered in deciding whether a warrant can be dispensed with, one being a “clear showing” of probable cause, id. at 392-93, does not suggest that all six factors must be present in each case; and a later District of Columbia Circuit decision, written by the author of Dorman, makes clear that all need not be. See United States v. Robinson, 533 F.2d 578, 583-84 (D.C.Cir.1976) (en banc). Moreover, in United States v. Acevedo, 627 F.2d 68, 70 (7th Cir.1980), we cautioned against the “checklist-type analysis” of Dorman (see also People v. Abney, 81 Ill.2d 159, 173, 41 Ill.Dec. 45, 51, 407 N.E.2d 543, 549 (1980); 1 LaFave & Israel, Criminal Procedure § 3.6, at pp. 262-63 (1984); LaFave, “Seizures” Typology: Classifying Detentions of the Person to Resolve Warrant, Grounds, and Search Issues, 17 J.L. Reform 417, 454-58 (1984)), and said the question was simply “whether the exceedingly strong privacy interest in one’s residence is outweighed by the risk that delay will engender injury, destruction of evidence, or escape.” The Fourth Amendment contains no checklist of factors constituting an emergency — contains, indeed, no reference to emergencies. The operative word in the Fourth Amendment is “unreasonable”; so the question ought to be, were the police unreasonable in not getting a warrant in the circumstances that confronted them?
The greater the danger to public safety if the police delay entering premises in search of a criminal suspect, the more reason they have for not waiting; and the danger here was greater than in Hayden, Dorman, or any other case we know of in which a “clear showing” of probable cause, as distinct from a mere showing, was made. The analogy to determining reasonableness in a negligence case by comparing the danger of an accident to the burden of avoiding it, see, e.g., United States v. Carroll Towing Co., 159 F.2d 169, 173 (2d Cir.1947) (L. Hand, J.), suggests that in determining whether police are reasonable in entering a house without a warrant the trier of fact ought to consider not only how great the risk of delay was — that is, the probability of injury, escape, or destruction of evidence, see, e.g., United States v. Acevedo, supra, 627 F.2d at 71 — but also how great the harm would have been had the risk materialized. The greater that harm would be, the less need be the probability that it would actually have occurred to justify the police in invading the interest (great though it is) in the privacy of the home. The potential harm from waiting for a search warrant in this case was very great even though it was far from certain that an immediate search would be productive. See United States v. Bottoson, 644 F.2d 1174, 1176 (5th Cir.1981) (per curiam); United States v. Jones, 635 F.2d 1357,1360 (8th Cir.1980); United States v. Williams, 612 F.2d 735, 739 (3d Cir.1979); United [1565]*1565States v. Bustamante-Gamez, 488 F.2d 4, 9 (9th Cir.1973).
Even so, the police could lawfully enter the Llaguno house without a warrant or the homeowner’s consent only if there was probable cause to believe that the killer was in the house. Although the words “probable cause” appear only in the second clause of the Fourth Amendment, which deals with warrants, and we have said that the police did not have to get a warrant in this case, the words are also used to describe an essential ingredient of reasonableness, see, e.g., Chambers v. Maroney, 399 U.S. 42, 51, 90 S.Ct. 1975, 1981, 26 L.Ed.2d 419 (1970); and all searches and seizures must be reasonable to comply with the Fourth Amendment.
It is true that the issue of probable cause ordinarily is for the judge rather than the jury. That is because the issue usually arises in the context of a motion to suppress evidence, which the judge decides. But where the issue arises in a damage suit, it is, as the panel opinion acknowledged, a proper issue for the jury if there is room for a difference of opinion. See 739 F.2d at 1190; Hindman v. City of Paris, 746 F.2d 1063, 1067 (5th Cir.1984); Garris v. Rowland, 678 F.2d 1264, 1270 (5th Cir.1982); Giordano v. Lee, 434 F.2d 1227, 1230 (8th Cir.1970); cf. Banish v. Locks, 414 F.2d 638, 641 (7th Cir.1969). The underlying issue in deciding whether the police had probable cause to do what they did is reasonableness, which is also the underlying issue in deciding negligence — a classic jury issue.
Probable cause means, in fact, a reasonable basis — “more than bare suspicion, but less than virtual certainty,” United States v. Garza-Hemandez, 623 F.2d 496, 499 (7th Cir.1980); see also Beck v. Ohio, 379 U.S. 89, 91, 85 S.Ct. 223, 225, 226, 13 L.Ed.2d 142 (1964); Grano, Probable Cause and Common Sense: A Reply to the Critics of Illinois v. Gates, 17 J.L. Reform 465, 478-512 (1984) — for believing that a search or seizure will be fruitful— will turn up evidence, or leads to evidence, or contraband, or the perpetrator of the crime. Emergency is not enough. An emergency, at least the kind of emergency, great but not apocalyptic in its menace, involved in this case, would not allow the police to search every house in Chicago or even every house on the Llagunos’ block. There must be something more than suspicion — there must be reasonable grounds for believing that the search of this house would prove fruitful in the criminal investigation. There is considerable doubt whether this requirement was met here. An example will dramatize our concern. A man drives to work, and parks his car on the street. A thief steals it, runs down a pedestrian, and crashes the car. The police check the license, note that the car has not been reported stolen, and, inferring from this that the car was being driven by its owner when it crashed, proceed forthwith to the owner’s home, sledgehammer in hand. They knock on the door and when no one answers they break it down, search the house, and arrest the owner when he returns from work. It is this kind of police, excess that the plaintiffs ask us to condemn by directing a verdict for them.
A reasonable jury would have to find a violation of the Fourth Amendment in our hypothetical case — to which the present case bears more than a family resemblance. Even the fact that a multiple murderer is on the loose does not give the police a license to search and seize without a reasonable basis, see, e.g., Lankford v. Gelston, 364 F.2d 197, 198 (4th Cir.1966) (en banc); 2 LaFave, Search and Seizure § 6.1, at p. 374 (1978), though it may affect the judgment of what is reasonable. But although the present case is close to the line that separates arguably reasonable from unarguably unreasonable police behavior, it does not cross it. Probable cause — the area between bare suspicion and virtual certainty — describes not a point but a zone, within which the graver the crime the more latitude the police must be allowed. The shooting of seven persons (four fatally) by a team of criminals in the space of two hours is about as grave a crisis as a local police department will en[1566]*1566counter. The police must be allowed more leeway in resolving it than when they are investigating the theft of a bicycle. Especially when a multiple murderer is at large in circumstances suggesting that he may be about to kill again, the interest in public safety is paramount.
It is true that the gravity of the crime and the threat of its imminent repetition usually are discussed in relation to the existence of an emergency justifying a search or arrest without a warrant, as we have already seen, rather than in relation to probable cause for the search or arrest. But there is some judicial recognition of the latter relation. See United States v. Preston, 468 F.2d 1007,1010 (6th Cir.1972); Nueslein v. District of Columbia, 115 F.2d 690, 696 (D.C.Cir.1940); Brinegar v. United States, 338 U.S. 160, 183, 69 S.Ct. 1302, 1314, 93 L.Ed. 1879 (1949) (Jackson, J., dissenting); cf. Grano, supra, 17 J.L. Reform at 503-04. The amount of information that prudent police will collect before deciding to make a search or an arrest, and hence the amount of probable cause they will have, is a function of the gravity of the crime, and especially the danger of its imminent repetition. If a multiple murderer is at large, the police must compress their investigation and make the decision to search or arrest on less information than if they could investigate at their leisure.
The police in this case had strong reason to believe that the killer who had fled on foot was Hispanic; and since the car had not been reported stolen and the registered owner had a Hispanic name, they had some reason to think the killer was either the owner or had been driving with the owner’s permission. Although the car was registered to “Vilma” Llaguno, a woman’s name — we now know — the police may not have known that Vilma is always a woman’s first name (a matter on which the record is unclear); readers of Chekhov know that “Vanya” is a man’s first name. Anyway, Vilma might have been the wife or sister or mother of the killer, in which event her address might have been his. And he might have sought refuge at her address (if she was a close relative) even if it was not also his, because the crash occurred on the major thoroughfare leading to the house (North Avenue), though not within easy walking distance (two miles). Of course the killer who fled on foot might not have been Vilma’s relative; the record does not reveal whether the police who entered the Llaguno home knew by then the name of the killer, Garcia, who had been seized at the crash. And if the fleeing killer was not a relative it was much less likely that he would flee to the home of the car’s owner. And yet we think a reasonable jury could have found that the police had a reasonable probability, based on real if inconclusive information rather than inspired hunch or a dragnet mentality, of finding the killer in (or en route to or from) the Llaguno home. As it turned out, he was not there; but that just shows that the probability he would be there was not 100 percent. The police were not completely off base. The killer was, after all, a son of the people who lived there, and he might well have fled there after the shooting spree, though in fact he did not.
Even so, it would not have been reasonable for the police to act on such limited information if they could have gotten better information first without incurring, or subjecting others to, great danger. There are several things they could have done. They could have tried to interview the occupants of the Llaguno home with the occupants’ consent. They could have sealed off the house and questioned the occupants as they emerged. They could have questioned neighbors. They could have waited till Garcia had recovered enough from his wound to be questioned. But each of these alternatives was dangerous. If the killer had actually been in the house, the policemen’s efforts to obtain the occupants’ consent to a search could have alerted him to the presence of the police and allowed him to shoot before they could disarm him. Sealing off the house may have been difficult, because it was nighttime and the house was a detached house which would have had to be surrounded. If the killer had been there, as the police had some [1567]*1567reason to think he was, sealing off the house might have led him to barricade it or take some of its occupants as hostages (people sometimes take their relatives hostage). Questioning Garcia, or the neighbors, could have caused a long delay in discovering whether the suspect was inside the house.
Given the gravity of the crimes they were investigating, the possibility that there would be more shootings unless the killer was seized immediately, and the information (limited as it was) that made it seem that he might well have fled to the Llaguno home, we cannot say, as a matter of law, that the police did not have probable cause to enter and search the house as they did. And if this is right, we do not think it makes a critical difference that they had no definite suspect in mind. If the police see a killer (or for that matter some lesser felon), whose name they do not know and whose features they cannot describe, enter a home, they can go in after him without violating the Fourth Amendment. See United States v. Santana, 427 U.S. 38, 42-43, 96 S.Ct. 2406, 2409-10, 49 L.Ed.2d 300 (1976). And their conduct is not unreasonable just because they are not 100 percent sure that this person is the killer they were after — that it was indeed he, and not some look-alike, who had fled into the house. See United States v. Stubblefield, 621 F.2d 980 (9th Cir.1980); United States v. Oaxaca, 569 F.2d 518, 520-22 (9th Cir.1978); United States v. Scott, 520 F.2d 697 (9th Cir.1975); United States v. Holland, 511 F.2d 38, 44-45 (6th Cir.1975); United States v. Shye, 492 F.2d 886 (6th Cir.1974) (per curiam). This case is more difficult than those we have cited because the police had less reason to be confident that they really were in hot pursuit of the killer when they entered the Llagunos’ house. But if we are right in our analysis of the situation that faced the police, there was a sufficient chain linking the house to the killer whom the police were pursuing to allow a reasonable jury to conclude that the police had probable cause to enter the house.
The search and seizure at the Llaguno home was over with when the police left a few hours after arriving. The 42-hour police detention of David Llaguno which ensued raises separate issues to which we now turn. If, as we believe the jury could reasonably have found in an error-free trial, the police were lawfully in the Llaguno house when they asked David to whom he had lent his car, then whether he refused to answer their question, as the police say, or answered with the name of his brother’s accomplice, as he says, the police (whether or not they knew that Garcia, the name David says he gave them, was the name of the' accomplice — a matter on which the record is unclear) had grounds for arresting him on suspicion that he was an accomplice in the crimes they were investigating. He had already acknowledged that the car used in the crimes was his. The discrepancy between the registration in Vilma Llaguno’s name and David Llaguno’s claim of ownership was not explored at trial; but the police were not required to ignore David’s claim of ownership just because the license-plate check had shown that the car was registered to someone else, though with the same last name. Cf. United States ex rel. Kirby v. Sturges, 510 F.2d 397, 401 (7th Cir.1975); 1 LaFave & Israel, supra, § 3.3, at pp. 211-12.
“[A] policeman’s on-the-scene assessment of probable cause provides legal justification for arresting a person suspected of crime, and for a brief period of detention to take the administrative steps incident to arrest. Once the suspect is in custody, however, the reasons that justify dispensing with the magistrate’s neutral judgment evaporate.” Gerstein v. Pugh, 420 U.S. 103, 113-14, 95 S.Ct. 854, 862-63, 43 L.Ed.2d 54 (1975). Although these words could be read to require that the arrested person be brought before a magistrate immediately upon arrival at the station house, instead they have been interpreted, consistently with language in the opinion contrasting “brief detention” after arrest, which is permitted, with “pro[1568]*1568longed” or “extended” detention, which is not, see 420 U.S. at 114, 95 S.Ct. at 863, to allow some interval for booking the arrested person and completing other paperwork before presenting him to the magistrate. See, e.g., Bernard v. City of Palo Alto, 699 F.2d 1023, 1024-25 (9th Cir.1983) (per curiam).
It did not require 42 hours to book David Llaguno and complete other paperwork necessary for bringing him before the magistrate to determine whether there was probable cause to hold him. Magistrates were available throughout the period, and assistant state’s attorneys twice told the police that there was insufficient evidence to place charges against David. The only reason for delay in bringing him before a magistrate was that the police hoped to build a case against David while he was in jail, and this is not a permissible reason for jailing someone indefinitely. It would inject an element alien to our system — imprisonment on suspicion, while the police look for evidence to confirm their suspicion. Of course the delay here was not indefinite, but it was almost two days; and with no better reason offered than that the police were still investigating David’s possible involvement in the crimes of his brother, it was too long.
David Llaguno was thus entitled to a directed verdict that the defendants were liable for his being held in jail — at least beyond the brief period that would have been necessary to book him and bring him before a magistrate. The qualification is necessary to take care of the possibility that the police were lawfully in the house when they questioned David, so that his arrest, at least, was lawful. As we are about to see, the issue of the lawfulness of the entry will have to be retried. If the entry is found to have been unlawful, the issue may arise whether the police can use information obtained as a result of the entry — the fruit of unlawful conduct — to justify the arrest of David in this suit. The question would not be whether such information had to be excluded from evidence, as it would be in a criminal trial of David, see Wong Sun v. United States, 371 U.S. 471, 83 S.Ct. 407, 9 L.Ed.2d 441 (1963). The applicability of the exclusionary rule in civil proceedings is a controversial issue, see, e.g., Immigration and Naturalization Service v. Lopez-Mendoza, — U.S. -, 104 S.Ct. 3479, 3485-90, 82 L.Ed.2d 778 (1984); Tirado v. Commissioner of Internal Revenue, 689 F.2d 307, 309-15 (2d Cir.1982), unnecessary to resolve here. The exclusionary rule does not come into play until a search is found to have been illegal; the rule provides a sanction for an illegal search. The' question here would be whether the arrest of David was legal because of what he told the police, even though they were able to question him only because they illegally entered the house where he lived. If he was coerced to respond to their questions, as he may very well have been, the information he gave could not be used to validate the arrest; but if he answered those questions voluntarily, it would be a jury question whether his decision to do so was an intervening cause which cut off the effect of the illegal entry.
The other plaintiffs, while not entitled to a directed verdict, are entitled to a new trial, because of cumulatively serious trial errors:
1. The entire instruction on probable cause was as follows: “[The jury must decide] whether the defendants reasonably believed that they had probable cause to enter the Llaguno house to arrest someone within it. What is a reasonable belief depends on the facts and circumstances within a defendant’s knowledge. Probable cause to arrest exists if a reasonable person would have believed that a crime had been committed and that a person within the house had committed the offense.” By defining “reasonable belief” solely in terms of the “facts and circumstances within a defendant’s knowledge,” the instruction deflects the reader (or hearer) from a central question: the reasonableness of the police in acting so hastily on the basis of their very limited knowledge, without investigating further. And the last sentence in the [1569]*1569instruction implies that if the police had reason to think there might be one criminal in the house, they were automatically entitled to round up all of the occupants of the house, including young children who could not possibly be criminals.
We add that, on retrial, the instructions should emphasize the importance that the Fourth Amendment has been interpreted to place on having a magistrate make the judgment of probable cause. The burden of proof should be placed on the police to establish the existence of an emergency that prevented them from obtaining a warrant.
2. The judge gave an instruction on immunity: “The law allows a defendant to defend a charge of unconstitutional entry by claiming a good faith belief that, under the circumstances, it was reasonable to enter the Llaguno house without a warrant. The defendant also must prove that such good faith belief was reasonable.” Harlow v. Fitzgerald,, 457 U.S. 800, 818-19, 102 S.Ct. 2727, 2738-39, 73 L.Ed.2d 396 (1982), decided before the trial of this case, held that the good-faith or qualified immunity of a civil rights defendant is not an issue for the jury, other than in exceptional circumstances not presented here. The issue of immunity as redefined in Harlow is whether when the defendant violated the plaintiff’s rights the law appeared to authorize the defendant’s misconduct. This is an issue of law, for the judge to decide. What the defendant believed, an issue of fact, is neither here nor there. The good sense of Harlow in withdrawing the issue of immunity from the jury is particularly evident in a case such as this, where the police are charged with having acted without probable cause. The question whether they had probable cause depends on what they reasonably believed with reference to the facts that confronted them, as the judge instructed in the passage we quoted earlier. To go on and instruct the jury further that even if the police acted without probable cause they should be exonerated if they reasonably (though erroneously) believed that they were acting reasonably is to confuse the jury and give the defendants two bites at the apple.
3 & 4. The judge told the jury that in evaluating the defendants’ behavior it should “not use 20/20 hindsight,” and “should consider the responsibility of the police to prevent crime, apprehend criminals, and to safeguard persons and property from criminal actions.” These instructions were not wrong in the sense of stating untruths, but they were gratuitous and prejudicial. The term “20/20 hindsight” is a derisory expression for an ex post facto judgment. The plaintiffs were entitled to ask the jury to make such a judgment, evaluating the conduct of the police long after the fact. And to remind the jury, quite unnecessarily one would have thought, that the police are responsible for protecting the public safety is to place the judge’s thumb on the balance in favor of a class of defendants already regarded sympathetically by most jurors. Cf. Briscoe v. LaHue, 460 U.S. 325, 342, 103 S.Ct. 1108, 1119, 75 L.Ed.2d 96 (1983); Darbin v. Nourse, 664 F.2d 1109, 1115 (9th Cir.1981).
5. Over the plaintiffs’ objections the judge allowed the defense to dwell in obsessive detail on the crimes that the police were investigating when they entered the Llagunos’ home — to bring out such facts as that the number of wounds of some of the victims and the caliber of the gun used on them could not be determined because the wounds were so “massive,” that a person had come running out of the tavern where the second robbery had taken place screaming “Oh God, they killed — they killed her, they killed her,” and that a victim’s “hand was cold and she was already starting to change color.” These details were not irrelevant, because they helped to show the danger that the police were up against. But dwelt on in such detail they may have turned the jurors against the plaintiffs — who are, after all, the killer’s close relatives, including his parents and brother. The judge should have found that the danger of unfair prejudice clearly outweighed the probative value of this evidence. Fed.R.Evid. 403.
[1570]*1570We need not decide whether any one of these errors would warrant reversal in and of itself, or even whether all together would warrant reversal in a different kind of case. But bearing in mind that civil rights actions often pit unsympathetic plaintiffs — criminals, or members of the criminal class (even — in this case — a multiple murderer’s parents and brother)— against the guardians of the community’s safety, yet serve an essential deterrent function especially at a time like this when the exclusionary rule is being narrowed (see United States v. Leon, — U.S.-, 104 S.Ct. 3405, 82 L.Ed.2d 677 (1984)), we take a serious view of trial errors that consistently favor the defendants in such a case. The fact that the jury brought in a judgment for the defendants even with respect to David Llaguno’s complaint of unlawful detention suggests that the jury was indeed swayed by the errors.
The appellants make other complaints about the conduct of the trial that are not well taken. But for the reasons we have explained, the district court’s judgment must be reversed and the case remanded with instructions to enter judgment for David Llaguno on Count VII and to grant the plaintiffs a new trial on all the other counts (plus a trial on Count VII limited to damages). Costs in this court to appellants.
Reversed and Remanded.