The issue on this appeal is whether placement of a prison inmate in administrative segregation in a Special Housing Unit (SHU) at the Auburn (N.Y.) Correctional Facility ("Auburn") for an aggregate interval that reached 101 days impairs a liberty interest for which procedural due process must be provided. The appeal also illustrates problems that can arise when liberty interest issues are litigated in a jury trial. Emmeth Sealey appeals from a judgment of the United States District Court for the Northern District of New York (David N. Hurd, Magistrate Judge), that gave judgment as a matter of law to Defendant-Appellee Lt. T.H. Giltner, an Auburn corrections officer, after a jury awarded Sealey $1 nominal damages. Applying the teaching of Sandin v. Conner, 515 U.S. 472 (1995), we conclude that the evidence was insufficient to support a finding that the 101-day confinement in the Auburn SHU impaired a protected liberty interest, and we therefore affirm.

Sealey's confinement in the SHU. Sealey has been an inmate in the New York state prison system since 1973. At all relevant times he was incarcerated at Auburn. On March 29, 1990, an Auburn inmate was slashed on his face and neck. Sealey was identified as having been involved in the incident. The next day, he was placed in administrative segregation within the Auburn SHU pending a disciplinary hearing on charges of violating rules against fighting, possession of a weapon, and assault. At a disciplinary hearing that concluded on April 9 (all dates are in 1990, unless otherwise noted), Sealey was found not guilty. The next day, while still in the SHU, he was served with an Administrative Segregation Recommendation, which stated:

Based on both confidential and other inf[ormation] on file in the DSS [Deputy Superintendent for Security] Office, it is felt that your continued presence in general population could seriously jeopardize the safety and security of this facility. The inf[ormation] on file indicates that you are involved in extortion and strong arm and you are being recommended for placement in Admin[istrative] Segregation.

A hearing on this Recommendation, required by prison regulations, see N.Y. Comp. Codes R. & Regs. tit. 7, 301.4(a) (1999), was held on April 16. Defendant-Appellee Lt. T.H. Giltner conducted the hearing. He denied Sealey's request for access to the confidential information on which the Recommendation had been made and refused Sealey's request to call witnesses. Giltner upheld the Recommendation that Sealey remain in administrative segregation because his "presence in general population would pose a threat to the safety and security of the facility." Id. 301.4(b).

On June 18, in an administrative appeal, Defendant Donald Selsky, acting director of special housing and inmate discipline, reversed Giltner's decision, on the ground that Giltner had failed to independently verify the confidential information and should have allowed Sealey to call witnesses.

On July 8, Defendant Lt. R. Brimmer conducted another hearing. Brimmer denied Sealey's request to call witnesses and determined, based on confidential information and Sealey's history of uncooperative behavior, that Sealey should remain in administrative segregation. Sealey's confinement in the SHU ended on August 29, when he was transferred to the Shawangunk Correctional Facility ("Shawangunk"), where he was released into general population. Thereafter, on September 7, in an administrative appeal, Selsky upheld Brimmer's decision, but on January 23, 1991, reversed Brimmer's decision.

Thus, Sealey was kept in the Auburn SHU for a period of 153 days, comprising five intervals: 11 days (March 30-April 9) prior to the end of the disciplinary hearing, 7 days (April 10-April 16) prior to the first administrative segregation hearing, 63 days (April 17-June 18) prior to the reversal of the first administrative segregation decision, 20 days (June 19-July 8) prior to the second administrative segregation hearing, and 52 days (July 9-August 29) from the second hearing to the transfer to Shawangunk.

Conditions in the SHU. Sealey and another witness testified at trial as to the conditions of confinement in the SHU. An inmate is confined to his cell 23 hours per day, can take no more than three showers per week, has limited library privileges and no telephone privileges.1 There was no quiet bell in the SHU, so it was noisy most of the time. On occasion, inmates threw feces at other inmates.

The pending litigation. On January 10, 1992, Sealey commenced this action, pursuant to 42 U.S.C. 1983, asserting a violation of his procedural due process rights. By agreement, the case was assigned to Magistrate Judge Hurd. He initially granted all four defendants' motions for summary judgment. On appeal to this Court, we affirmed the dismissal as to Defendant Coughlin for lack of personal involvement, and reversed and remanded with respect to Giltner, Brimmer, and Selsky in order to permit development of the facts as to whether the conditions of Sealey's confinement in the SHU were "atypical" within the meaning of Sandin. See Sealey v. Giltner, 116 F.3d 47, 49 (2d Cir. 1997) ("Sealey I").

The trial. Magistrate Judge Hurd tried the case with a jury in November 1997. At the close of Sealey's evidence, the Defendants moved for judgment as a matter of law. Magistrate Judge Hurd reserved decision on their claim that the evidence did not show a deprivation significant enough to constitute a liberty interest, denied their claim that the evidence did not show lack of procedural due process, and dismissed Sealey's claim against Selsky for lack of a showing that any act by Selsky caused harm to Sealey (Selsky's affirmance of Brimmer's decision to continue Sealey's confinement did not occur until after Sealey was transferred).

After the close of all the evidence, the remaining Defendants, Giltner and Brimmer, renewed their motions to dismiss. The Court responded:

With regard to the motion to dismiss on the basis that the plaintiff was not deprived of a liberty interest, the liberty interest question is a matter of law which will be decided by me. If this was a bench trial, the Court would be inclined to make that decision now and perhaps rule that plaintiff does not have a liberty interest and proceed to dismiss the case in its entirety. However, we do have a jury and the plaintiff has raised some issues to demonstrate that his confinement in SHU was [] atypical. He was confined for 145 days.2 The confinement was indefinite duration, which could have lasted much longer and was only ended because he was transferred to another facility. He has testified about feces being thrown, about the excess noise, and the other matters concerning his SHU confinement. Therefore, the question is not clear-cut that he did not have a liberty interest.

The Magistrate Judge added that he was reserving the issue and putting the question to the jury so that there would be both a jury verdict and his decision on the liberty interest to review on appeal. The Court dismissed as a matter of law Sealey's claim for confinement for the intervals while he was awaiting the disciplinary hearing and awaiting the first administrative hearing.

Magistrate Judge Hurd instructed the jury that it was to consider only the SHU confinement between April 16, 1990, the date of the first administrative hearing, and Sealey's transfer to another facility. He also provided the jury with a verdict form that asked several questions. Recognizing Sealey's claim that the SHU confinement, though labeled administrative, was in reality disciplinary, the Court instructed the jury on the difference between the two types of confinement and explained the due process procedural requirements of each type.

Though the Magistrate Judge had said that the facts concerning the deprivation of a liberty interest were not "clear-cut," and that it would be helpful on review to have a jury verdict on the issue in the event that this Court disagreed with any ruling of law he might make, he did not instruct the jury on what constitutes a liberty interest in the prison context, nor frame a jury question asking whether Sealey had been denied a liberty interest. However, the Court alluded to the liberty interest issue in its instruction on actual damages, stating that, in determining the amount of any actual damages, the jury "may take into consideration the [effect] that Mr. Sealey's confinement in SHU as opposed to the general prison population had on his ability to enjoy life" and that the amount of any actual damages awarded "would be the difference, if any, between [SHU] confinement and the general prison population as it affects the plaintiff." Sealey's counsel expressed only "a general objection" to the charge, explicitly declining to make any specific objections.

The jury determined that Sealey's confinement was administrative, that Giltner (but not Brimmer) denied Sealey procedural due process, and that the violation was not the proximate cause of any actual damages. The jury awarded Sealey $1 in nominal damages.

Post-trial rulings. After the verdict, Giltner renewed his motion for judgment as a matter of law, and the District Court granted the motion. The Court determined that under Sandin Sealey "failed to factually demonstrate that his administrative confinement in SHU was an atypical and significant hardship." Sealey v. Coughlin, 997 F. Supp. 316, 320 (N.D.N.Y. 1998). Explaining this conclusion, Magistrate Judge Hurd stated:

Other than the plaintiff's own testimony that he had feces thrown in his face and was unable to concentrate and sleep because of constant noise in SHU, there was no other evidence at trial to support the contention that his confinement in administrative segregation was atypical and a significant hardship. No records, reports, or complaints were filed with the Auburn Correctional Facility verifying plaintiff's allegations of inhuman treatment. Consequently, plaintiff's allegations are unsupported conclusions. Plaintiff's testimony alone, without corroborating evidence, cannot establish a liberty interest. To rule otherwise would allow a plaintiff/inmate to meet the "atypical" hardship threshold by merely testifying about some brutal conditions or treatment in SHU.

Id. at 321 (citation omitted). The Court concluded that the conditions of the SHU are the sort of confinement that Sealey should reasonably anticipate receiving at some point in his incarceration. The Court added that the jury's finding of no actual damages was evidence that Sealey had not suffered any deprivation.

Finally, the Court denied Sealey's application for attorney's fees under 1988.

The Supreme Court's decision in Sandin made clear that a prisoner's restricted confinement within a prison does not give rise to a liberty interest, warranting procedural due process protection, unless the conditions and duration of the prisoner's confinement "impose[] atypical and significant hardship on the inmate in relation to the ordinary incidents of prison life." 515 U.S. at 484. For convenience, we will use the word "atypical" to describe restricted confinement that imposes both atypical and significant hardship. The Sandin ruling has generated several issues. Among them are: (1) Does a liberty interest arise whenever a prisoner is placed in "atypical" confinement only for disciplinary reasons or also for administrative reasons? (2) Does a liberty interest arise only when a state specifies that a specific factual predicate must exist before placement in "atypical" confinement may occur or also when such placement is entirely discretionary? (3) In determining atypicality, is the relevant comparison with conditions of the general prison population, or those in administrative confinement, or both? (4) Is atypicality affected by the duration of the challenged confinement? (5) Is atypicality affected by the frequency of confinements of a duration at least as long as the challenged confinement? (6) Whichever comparative conditions are relevant and if frequency of confinements of at least equal duration is relevant, are the comparative conditions and durations those of the prison in which the inmate is confined, of all prisons within the same state, or of all prisons in the nation? (7) Is atypicality affected by the length of the prisoner's sentence, such that conditions might be deemed atypical for a prisoner serving a short sentence but not a long sentence? (8) Is atypicality to be determined prospectively or retrospectively? (9) Where different prison officials have responsibility for a prisoner's confinement during different segments of an aggregate time period that constitutes "atypical" confinement, for which segments are the officials required to observe procedural due process requirements? Fortunately, we need not consider all of these issues to resolve the pending appeal, but some of them are before us, and the identification of others might assist those encountering similar cases in the future.

Whatever Sandin tells us is best understood in the context of what preceded that decision. In Hewitt v. Helms, 459 U.S. 460 (1983), the Court considered a procedural due process claim by a prisoner placed in restricted confinement for administrative reasons. The Court first considered whether the confinement implicated a liberty interest "independently protected by the Due Process Clause." Id. at 468. Noting that "the transfer of an inmate to less amenable and more restrictive quarters for nonpunitive reasons is well within the terms of confinement ordinarily contemplated by a prison sentence," id., the Court rejected a liberty interest based solely on the Due Process Clause. Nevertheless, the Court ruled that, because state law had established "specific substantive predicates," id. at 472, that had to be found before restrictive confinement could be imposed, the prisoner had a protected liberty interest in avoiding such confinement.

In Sandin, the Court revisited Hewitt and rejected much of its reasoning. Specifically, the Court rejected the idea that a state's establishment of a specific substantive predicate for restrictive confinement of a prisoner is sufficient to create a protected liberty interest. Instead, though "recogniz[ing] that States may under certain circumstances create liberty interests which are protected by the Due Process Clause," Sandin, 515 U.S. at 483-84, the Court ruled that "these interests will be generally limited to freedom from restraint which ... imposes atypical and significant hardship on the inmate in relation to the ordinary incidents of prison life," id. at 484.

It is arguable that Sandin abolished the inquiry as to whether a state has established a substantive factual predicate for restrictive confinement and limited the inquiry to whether the conditions of confinement are atypical. The Chief Justice's opinion for the Court pointed out that the substantive factual predicate inquiry of Helms creates a disincentive for states to codify their prison management procedures and leads to extensive involvement of federal courts in matters of day-to-day prison management. See id. at 482-83. Moreover, the decision states, albeit in a footnote, that it "abandons an approach that in practice is difficult to administer and which produces anomalous results." Id. at 483 n.5 (emphasis added). However, the Court's recognition that the states can create a protected liberty interest in freedom from restricted confinement (such as SHU) must mean that the Hewitt inquiry as to a substantive factual predicate has not been completely abandoned, but only limited to those instances where a finding of such a predicate results in confinement to conditions of atypical and significant hardship. That is how Sandin has been understood in this Circuit. See Welch v. Bartlett, 196 F.3d 389, 394 n.4 (2d Cir. 1999) ("Hewitt, as modified by Sandin's additional requirement of an 'atypical and significant hardship,' remains good law in analyzing whether statutory or regulatory tests can generate a liberty interest."); Sealey I, 116 F.3d at 52; Frazier v. Coughlin, 81 F.3d 313, 317 (2d Cir. 1996).

Also arguable from Sandin is the proposition that restricted confinement for administrative reasons can never implicate a protected liberty interest. As we have observed, "Sandin may be read as calling into question the continuing viability of our cases holding that New York regulations afford inmates a liberty interest in remaining free from administrative segregation." Rodriguez v. Phillips, 66 F.3d 470, 480 (2d Cir. 1995).3 This argument draws support from Sandin's requirement of conditions of atypical and significant hardship, coupled with its recalling that in Hewitt the Court had "concluded that the transfer to less amenable quarters for nonpunitive reasons was 'ordinarily contemplated by a prison sentence.'" Sandin, 515 U.S. at 480 (quoting Hewitt, 459 U.S. at 468). Moreover, in Sandin, the prisoner's confinement was not deemed atypical, in part, because his "disciplinary segregation, with insignificant exceptions, mirrored those conditions imposed upon inmates in administrative segregation and protective custody." Id. at 486 (footnote omitted); see id. at 489 n.1 (Ginsburg, J., with whom Stevens, J., joins, dissenting) ("The Court reasons that Conner's disciplinary confinement, '... mirrored ... administrative segregation ...,' and therefore implicated no constitutional liberty interest.") (emphasis added and citation omitted) (quoting majority opinion at 486). The Seventh Circuit has interpreted Sandin to mean that "the key comparison [to determine atypical and significant hardship] is between disciplinary segregation and nondisciplinary segregation," Wagner v. Hanks, 128 F.3d 1173, 1175 (7th Cir. 1997), thereby implying that confinement for administrative reasons can never implicate a liberty interest.

We think Sandin does not go so far. The Court might have assumed that administrative confinement sometimes will not implicate a liberty interest because it might be imposed without the requirement that corrections officers find a substantive factual predicate. Thus, in one passage, the Court compared the conditions of Conner's disciplinary confinement to "similar, but totally discretionary, confinement." Sandin, 515 U.S. at 486 (emphasis added). But, as we have long recognized, New York has established substantive factual predicates for many instances of administrative confinement. See N.Y. Comp. Codes R. & Regs. tit. 7, 301.1-.7 (1999); Arce v. Walker, 139 F.3d 329, 334 (2d Cir. 1998) (applying Sandin's two-part inquiry, concerning state-created liberty interest and atypical restricted confinement, to prisoner held in administrative segregation); Brooks v. DiFasi, 112 F.3d 46, 49 (2d Cir. 1997) ("[W]e have never held that New York prisoners have no liberty interest in avoiding long-term administrative confinement."). New York's regulations specify that "[a]dministrative segregation admission [the form of administrative confinement imposed on Sealey] results from a determination by the facility that the inmates' presence in general population would pose a threat to the safety and security of the facility." N.Y. Comp. Codes R. & Regs. tit. 7 301.4(b) (1999). Moreover, after the Supreme Court acknowledged that a liberty interest exists when state law requires a substantive factual predicate for atypical restricted confinement, it is difficult to see why such an interest should not arise when such confinement is imposed for administrative reasons after determination that a required factual predicate has been established. If an inmate is to be placed in atypical confinement (considering both the conditions and the duration) after being determined, for example, to be a threat to prison safety, he should have some procedural due process surrounding the determination that he poses such a threat. That is the teaching of Hewitt, and if Sandin had meant to overrule Hewitt to the extent of precluding a protected liberty interest for all administrative confinements, we would expect to see more pointed language to that effect.

Since Sealey's administrative confinement occurred after a required factual determination that he posed a threat to prison safety, we must consider whether he was confined to conditions of atypical and significant hardship. We therefore turn to the determination of atypicality.

Initially, we pause to consider the procedural issue of how the issue of Sandin atypicality should be handled in a jury trial. The Magistrate Judge stated that he would decide the issue as a matter of law, but would submit the issue to the jury as a precaution so that there would be a verdict in the event that his ruling as a matter of law was rejected on appeal. However, his proposed procedure was not completely implemented. The jury charge did not instruct the jury that an element of the Defendants' liability was a finding that the conditions of Sealey's confinement were atypical, and the verdict form did not frame a question on this issue. Instead, the issue of atypicality was included, somewhat obliquely, as an aspect of actual damages, with very little indication of the appropriate standard.

We agree with the Magistrate Judge that the ultimate issue of atypicality is one of law, but that does not always mean that it need not be submitted to the jury. The content of the Sandin standard of "atypical and significant hardship" is an issue of law, but if the facts concerning the conditions or the duration of confinement are reasonably in dispute, the jury (where one is claimed) must resolve those disputes and then apply the law of atypicality, as instructed by the Court. In appropriate cases, the trial court might consider submitting interrogatories to the jury and then itself applying the law of atypicality to the facts as found by the jury. In this case, some of the facts about the conditions in the Auburn SHU were in dispute, but the atypicality issue was not explicitly submitted to the jury. The verdict form asked whether the confinement was administrative; if so, whether Giltner denied Sealey procedural due process at the April 16, 1990, hearing; and if so, whether the "violation" was a proximate cause of any actual damages. But there can be no "violation" of procedural due process requirements unless the confinement meets the atypicality standard of Sandin.4

Nevertheless, the Magistrate Judge's failure to submit the atypicality issue to the jury as an element of the claimed procedural due process violation does not warrant any relief. The Plaintiff made no objection to a verdict form that framed specific questions for the jury, and Rule 49(a) of the Federal Rules of Civil Procedure specifies that where a special verdict form is used, the omission of any issue waives the right to a jury trial on that issue and permits the court to make a finding. See Goeken v. Kay, 751 F.2d 469, 474 (1st Cir. 1985); see also Getty Petroleum Corp. v. Island Transportation Corp., 878 F.2d 650, 656 (2d Cir. 1989) (issue omitted from special verdict form implicitly decided by trial court). We therefore turn next to the Magistrate Judge's finding that Sealey's confinement was not atypical.

In finding that Sealey's confinement was not atypical, the Magistrate Judge erred in his intermediate ruling concerning Sealey's testimony, though not in his ultimate conclusion. After noting that Sealey's claim of atypicality was not supported by prison reports or other prisoners' complaints, the Magistrate Judge ruled that "Plaintiff's testimony alone, without corroborating evidence, cannot establish a liberty interest. To rule otherwise would allow a plaintiff/inmate to meet the 'atypical' hardship threshold by merely testifying about some brutal conditions or treatment in SHU." Sealey, 997 F. Supp. at 321. There is no basis for such a rule, and we decline to adopt it. Triers of fact, whether juries or judges, have the obligation to assess the credibility of witnesses, including parties, and a plaintiff's testimony in support of his own claim, if credited and persuasive, will normally suffice. This does not entitle a plaintiff to win merely by testifying, as the Magistrate Judge apprehended, but only by presenting testimony that the trier concludes is both credible and, along with other evidence, if any, sufficiently persuasive to satisfy the plaintiff's burden of proof by the requisite preponderance of the evidence.

The unwarranted rejection of Sealey's testimony, without regard to its credibility, does not necessarily warrant further fact-finding, however, unless that testimony, even if credited, would establish confinement in such conditions and for such duration as to satisfy the Sandin standard of atypicality. Both the conditions and their duration must be considered, see Welch v. Bartlett, 196 F.3d at 392-93, since especially harsh conditions endured for a brief interval and somewhat harsh conditions endured for a prolonged interval might both be atypical.

In assessing whether Sealey's confinement was atypical, we are hampered by a less than adequate record, despite our prior remand for development of the relevant facts. For example, there is no indication whether SHU confinement in conditions similar to Sealey's is imposed for administrative reasons in the complete discretion of prison officials, or, if so, how frequently that occurs and for what duration. Nor is there any evidence as to how frequently SHU confinement is imposed for administrative reasons after establishment of a required factual predicate and for what duration. See Brooks, 112 F.3d at 49 ("[T]he mere fact that New York's prison regulations permit extended administrative segregation does not tell how frequently or for what durations such segregation is imposed."). The absence of these and other facts necessitated a remand in Welch and in Brooks, where the prisoner's claim was rejected on motion for summary judgment. However, the pending case having been tried (partly to a jury and partly to the Magistrate Judge), the Plaintiff can prevail only if the evidence of record is legally sufficient to support his claim.

Sealey testified that he was confined to his SHU cell for 23 hours a day with one hour out for recreation, he was limited to three showers per week, and he lost various privileges. He described the SHU cells as noisy, much more so than general population cells, where, he testified, a "quiet bell" ends the time each day when a prisoner is allowed to "talk out of [his] cell." He also testified that "a few times" other SHU inmates threw feces at him.

With regard to the durational aspect of the atypicality issue, we must focus only on the interval during which Defendant Giltner is responsible, since, on this appeal, it is his alleged denial of procedural due process for which Sealey seeks reinstatement of the jury's award of nominal damages. The hearing that Giltner conducted was held on April 16, 1990. His decision ordering administrative segregation was reversed on June 18, 63 days later. The second hearing (not conducted by Giltner) was held on July 8, 20 days later.5 Thus, the longest interval following Giltner's hearing, for which he could be responsible, totals 83 days.6 However, Sealey had been in SHU segregation for 18 days before Giltner's hearing--11 days prior to the end of the disciplinary hearing and7 days prior to Giltner's administrative hearing. Although the confinement for these 18 days alone is not alleged to have implicated a liberty interest and although Giltner had no responsibility for determining the procedures (or their lack) with respect to such confinement, he nonetheless bears responsibility for the 101-day aggregate of the 83 days of confinement after his hearing and the 18 days that preceded it. Wherever the point is beyond which confinement in harsh conditions constitutes atypicality, a prison official must not be permitted to extend such confinement beyond that point without according procedural due process. 7 Prospectively, the entire interval of confinement in harsh conditions prior to a hearing will inform the official's decision whether to accord procedural due process when finding a factual predicate that state law requires as a condition of restrictive confinement. See Sandin, 515 U.S. at 489 n.1 (Ginsburg, J., dissenting) ("One must, of course, know at the start the character of the interest at stake in order to determine then what process, if any, is constitutionally due.") (emphasis in original). Retrospectively, that interval, plus the interval for which the official is responsible after the hearing, will inform a court's decision whether a prisoner confined in harsh conditions for the aggregate of both intervals is entitled to any damages.

Having examined the conditions and the duration of Sealey's confinement, we next consider the base against which to make the Sandin comparison of atypicality. Three subsidiary issues arise. First, to what type of confinement is the challenged confinement to be compared? Second, whichever type of confinement is relevant, is the comparison to conditions at the inmate's prison, all prisons in the state system, or all prisons in the nation? Third, if the conditions of confinement and their duration, in light of the appropriate comparison, are deemed to impose a significant hardship, what base is to be used in determining whether the hardship is imposed so infrequently as to render the challenged confinement atypical?

Uncertainty concerning the first issue stems from ambiguities in the Supreme Court's opinion in Sandin. Though that opinion stated generally that restricted confinement impairing a liberty interest must impose an "atypical and significant hardship" compared to "the ordinary incidents of prison life," 515 U.S. at 484, it did not rule definitively on the type of conditions that should be considered to constitute such "ordinary incidents." In fact, the opinion offered three different possibilities. At one point, the opinion noted that the conditions of Conner's disciplinary confinement "mirrored those conditions ... in administrative segregation and protective custody." Id. at 486. Next, the opinion observed that Conner's confinement "did not exceed similar, but totally discretionary, confinement in either duration or degree of restriction." Id. Though this language might be read to mean that the relevant comparison is only with restricted confinement imposed in the complete discretion of prison officials, the District of Columbia Circuit has pointed out that the Supreme Court very likely did not mean such a limitation. In Hatch v. District of Columbia, 184 F.3d 846 (D.C. Cir. 1999), that Court observed that the Sandin opinion referred to applicable Hawaiian regulations that circumscribed prison officials' authority to impose administrative segregation. Id. at 855-56 (citing Haw. Admin. Rule 17-201-22, 17-201-23). The panel in Hatch concluded that Sandin's "use of the words 'totally discretionary' cannot mean that what prison officials may do in their unfettered discretion is the touchstone for elucidating 'the ordinary incidents of prison life.'" Id. at 855. Finally, Sandin provides some indication that conditions in general population are relevant since it noted that "the conditions at Halawa [where Conner was confined] involve significant amounts of 'lockdown time' even for inmates in the general population." Sandin, 515 U.S. at 486 (emphasis added). A footnote to that sentence adds, "General population inmates are confined to cells for anywhere between 12 and 16 hours a day, depending on their classification." Id. at 486 n.8 (emphasis added).

These various comments in Sandin have led to differing views among the federal appellate courts. The Ninth Circuit compares restricted confinement to conditions for the general prison population. See Keenan v. Hall, 83 F.3d 1083, 1089 (9th Cir. 1996), reh'g denied and amended by 135 F.3d 1318 (9th Cir. 1998). The Fourth Circuit also appears to do so, see Beverati v. Smith, 120 F.3d 500, 504 (4th Cir. 1997), although it concluded that six months of confinement in especially disgusting conditions that were "more burdensome than those imposed on the general prison population" were not "atypical ... in relation to the ordinary incidents of prison life." Id. In dealing with challenges to disciplinary confinement, the Third and Seventh Circuits compare to conditions in non-disciplinary confinement, such as administrative segregation and protective custody, see Wagner, 128 F.3d at 1175; Griffin v. Vaughn, 112 F.3d 703, 707-08 (3d Cir. 1997), thereby ruling by implication that administrative confinement itself does not impair a liberty interest. The District of Columbia Circuit uses the "most restrictive" conditions of administrative confinement as the base for comparison of conditions of disciplinary confinement, but leaves open the possibility that administrative confinement itself might impair a liberty interest, since its comparison is to administrative confinement that prison officials "routinely impose on inmates serving similar sentences." Hatch, 184 F.3d at 856. The implication is that an inmate in administrative confinement could establish impairment of a liberty interest if he was confined in administrative confinement under conditions or for a duration more adverse than those "routinely" imposed on inmates serving similar sentences. The Fifth Circuit has ruled that disciplinary segregation (and a fortiori administrative confinement) does not impair a liberty interest unless the confinement status inevitably affects the duration of the sentence, see Carson v. Johnson, 112 F.3d 818, 821 (5th Cir. 1997), as might occur, for example, if restrictive confinement automatically forfeited good time credits.

The relevant comparison in this Circuit has not been definitively settled, although our decision in Brooks suggests that, in a disciplinary confinement case, a comparison might be made to both conditions in administrative confinement and in the general prison population. See Brooks, 112 F.3d at 48 (rejecting summary judgment for prison officials for lack of any findings "about the prevailing conditions in administrative confinement or in the prison at large"). Welch points in the same direction. See Welch, 196 F.3d at 394-95.

On the second issue (whether to compare to the inmate's prison or other prisons), the only Circuit to have explicitly considered the issue is the Seventh Circuit, which has ruled that the comparison is with other prisons in the state where the plaintiff is confined, unless the inmate's prison is the most secure in the state. See Wagner, 128 F.3d at 1175.

On the third issue (the relevant base for determining whether the conditions and durations of confinement that constitute severe hardship are imposed with such infrequency as to render the challenged confinement atypical), courts have provided little illumination, though the significance of frequency has been noted. See Welch, F.3d at 393-94; Hatch, 184 F.3d at 857; Brooks, 112 F.3d at 49.

In the pending case, the Plaintiff's and the Defendants' evidence (a) compared Sealey's SHU administrative confinement to conditions of general population inmates, (b) provided comparative information only for general population inmates at Auburn, and (c) presented no data concerning the frequency of confinement for durations as long as Sealey's. Since this is the record made after an opportunity for a full trial, we are limited to it and will therefore compare Sealey's SHU confinement to the conditions applicable to the general population at Auburn. We acknowledge, however, that conditions of administrative confinement at other New York prisons, as well as the frequency and duration of confinements imposing significant hardships, might well be relevant to an inmate's liberty claim.

Assessing Sealey's 101-day confinement in the Auburn SHU (for which Giltner bears responsibility) in light of these principles and with his testimony credited, we agree with the Magistrate Judge that he has not shown confinement of a duration and in such conditions as to meet the Sandin standard of atypicality. The conditions of the Auburn SHU are doubtless unpleasant and somewhat more severe than those of general population, but the degree of incremental harshness, endured for 101 days, is not an "atypical and significant hardship on the inmate in relation to the ordinary incidents of prison life." Sandin, 515 U.S. at 484.

The judgment of the District Court is affirmed.