POSNER, Chief Judge.

A state prisoner seeks federal habeas corpus, claiming to have been deprived of liberty within the meaning of the due process clause of the Fourteenth Amendment by being placed in disciplinary segregation. The principal though not only question presented by his appeal from the dismissal of his claim is whether, in determining whether such a deprivation has occurred, the district court must compare conditions in disciplinary segregation with those in which the general population of the prison is confined or with those in which the general population of any prison in the state is confined.

The question arose less frequently before Sandin v. Conner, 515 U.S. 472 (1995). Before then the deprivation of any right that the state conferred on a prisoner to some vestige or modicum of freedom within the prison was actionable, so that when a prisoner's status within the prison system changed the question was not how great the change was but whether it infringed a right that the state had conferred on him. See, e.g., Wallace v. Robinson, 940 F.2d 243 (7th Cir. 1991) (en banc). Sandin shifted the focus from whether there was an entitlement to whether the entitlement was to some meaningful amount of liberty. Conner, the prisoner in that case, had been ordered to spend thirty days in disciplinary segregation, which generally means confinement to one's cell for at least 23 hours of the day. Oddly, none of the opinions in Sandin indicates what being in "disciplinary segregation" means, concretely; the 23-24 hour figure is from the state's brief. All that seemed important to the Court was that conditions in disciplinary segregation were essentially the same as those in administrative segregation — which is to say segregation ordered for nondisciplinary reasons, such as that the prisoner is awaiting classification or transfer, or is an escape risk, or is incorrigible, or is a gang leader, or has a contagious disease — and in protective custody, which is also segregation, but is normally though not always requested by the inmate himself and the usual purpose of which is to protect the prisoner from being attacked by other prisoners. Conner might easily have found himself in either situation, that is, in administrative segregation or protective custody, and with no remedy, even if he had not misbehaved. Since he thus had no right to a more extensive liberty than that conferred by disciplinary segregation, he had not been deprived of liberty. "Conner's discipline in segregated confinement did not present the type of atypical, significant deprivation in which a state might conceivably create a liberty interest." 515 U.S. at 486.

Our prisoner, Wagner, was ordered to serve a year in disciplinary segregation in the Wabash Valley Correctional Institution, in Indiana, as punishment for committing a battery. He seeks federal habeas corpus, claiming that he had been denied due process of law in the proceeding that resulted in this sanction. The district judge dismissed the suit, citing Sandin, but without elaboration. The record is limited to the complaint, and contains no information about the relation between the conditions of disciplinary segregation in the Wabash Valley Correctional Institution and the conditions in which the rest of the prison's population is confined, the conditions in which prisoners in administrative segregation or protective custody are confined in that prison, or the conditions of confinement of inmates of other Indiana prisons. The state has told us that Indiana has some "Level V" prisons in which the entire inmate population is confined to its cells for 23 hours of the day. It has not told us whether the Wabash Valley Correctional Institution is one of these, but in its brief in another habeas corpus case brought by Wagner (a case that arose out of another battery, resulted in an identical punishment, and is being decided in an unpublished opinion issued today), it tells us that the Wabash Valley Correctional Institution has both Level III and Level IV facilities. But it does not tell us how restrictive these are, or which one Wagner is confined in, or how the conditions in the disciplinary unit compare with conditions either of the general population of either facility or of those prisoners who are in nondisciplinary forms of segregation. So, at least if the proper comparison is intraprison rather than interprison (an important qualification, and one that we shall relax shortly), the case has to be remanded. It is true that the complaint does not allege that the conditions of disciplinary segregation at Wabash are "atypical," but it does allege a "loss of liberty," and no more is required to satisfy the requirements of notice pleading under the Federal Rules of Civil Procedure.

We want to give the district court as much guidance as possible, both for this and future cases. So we begin by noting that if Wabash were a Level V prison, it would be highly unlikely that Wagner could satisfy Sandin's test. The reason is not that disciplinary segregation could not in principle be made more restrictive than even the virtually solitary confinement that one finds in Level V prisons and their counterparts in other states and in the federal prison system; for even maximum-security prisons have segregation units. E.g., Gometz v. Henman, 807 F.2d 113, 114 (7th Cir. 1986). But it would be difficult (we do not say impossible) to make disciplinary segregation sufficiently more restrictive than the conditions of the general population of such a prison to count as an atypical and significant deprivation of liberty — that is, to count as a substantial incremental deprivation — without scraping up against the Eighth Amendment's prohibition against cruel and unusual punishments.

Since Wabash is not a Level V prison, it is more likely that confinement in disciplinary segregation constituted a substantial incremental deprivation of Wagner's liberty. More likely, but not very likely, because the prison is likely to provide facilities for, and create conditions of, administrative segregation and protective custody that are virtually identical to the facilities for and conditions of disciplinary segregation, and no more is necessary under Sandin to deny the prisoner's claim. The manual of policies and procedures issued by the Indiana Department of Corrections indicates that the facilities and conditions are indeed the same in disciplinary and nondisciplinary segregation except that prisoners in administrative segregation or protective custody may be permitted "contact" visits and are entitled to make phone calls to persons other than lawyers. The denial of so limited an increment of privileges would be unlikely to effect a significant deprivation of liberty, but we hesitate to base decision on a procedures manual that may not be accurate or up to date. Cf. Brooks v. DiFasi, 112 F.3d 46, 49 (2d Cir. 1997).

We add for completeness that if Wabash were one of those "country club" prisons in which the prisoners enjoy a great deal of freedom, spending little time in their cells or even in the prison buildings, confinement for a protracted period in what amounts to solitary confinement would indeed work an atypical and significant deprivation of Wagner's liberty. See, e.g., Greaves v. New York, 951 F. Supp. 33, 36 (S.D.N.Y. 1996). But that, like the rest of our discussion so far, is on the assumption that the proper comparison group is the population of the petitioner's own prison, and it may not be; it may be any prison population in the state (or perhaps in the entire country, but we'll defer that issue for a bit). In that event Wagner must lose unless the conditions of disciplinary segregation in Wabash Valley Correctional Institution are substantially more restrictive than even those in the state's Level V prisons. This is conceivable, since, as we noted, even maximum-security prisons have segregation units. But it is improbable, for remember that under Sandin the key comparison is between disciplinary segregation and nondisciplinary segregation rather than between disciplinary segregation and the general prison population.

We do not think that comparison can be limited to conditions in the same prison, unless it's the state's most secure one. Griffin v. Vaughn, 112 F.3d 703, 708-09 (3d Cir. 1997). (Cf. Dominique v. Weld, 73 F.3d 1156, 1159-61 (1st Cir. 1996), holding that the proper comparison is to the conditions of "ordinary prison life" rather than to the conditions in which the particular prisoner was being held prior to his disciplinary sanction.) To distinguish between the different parts of the same prison, on the one hand, and the different prisons in the same system, on the other, would be arbitrary. The decision whether to create high-security segregation units in each prison or to concentrate them in one or a few special prisons bears no relation to any interest protected by the Constitution. It is no surprise, therefore, that the courts have held that the transfer of a prisoner from one prison to another is not actionable as a deprivation of constitutionally protected liberty even if the conditions of confinement are much more restrictive in the prison to which the prisoner is being transferred. Meachum v. Fano, 427 U.S. 215 (1976); Montanye v. Haymes, 427 U.S. 236 (1976); Olim v. Wakinekona, 461 U.S. 238 (1983); see Bryan v. Duckworth, 88 F.3d 431, 434 (7th Cir. 1996). To have held otherwise would as a matter of logic have required the courts to adjudicate transfers within a prison — to determine, for example, whether the petitioner had been deprived of liberty by being transferred from a large cell to a small one. Federal judges would have been plunged deep into the minutiae of prison administration, much as if they were managing a hotel chain.

When Sandin is interpreted in light of the transfer cases, it becomes apparent that the right to litigate disciplinary confinements has become vanishingly small. Cf. Beverati v. Smith, 120 F.3d 500, 504 (4th Cir. 1997). This may be why, since Sandin, no court of appeals has found such a confinement to be a deprivation of liberty, though Williams v. Fountain, 77 F.3d 372, 374 n. 3 (11th Cir. 1996), "assumed" a deprivation, and the Second Circuit seems inclined to leave the door open a bit wider than Sandin, when read in light of the transfer cases, appears to allow. See Brooks v. DiFasi, supra; Miller v. Selsky, 111 F.3d 7, 9 (2d Cir. 1997). This is not to say that prison discipline is no longer actionable in the federal courts. When it takes the form of prolonging the prisoner's incarceration or otherwise depriving him of what has been held to be liberty or property within the meaning of the due process clauses, it is securely actionable. See, e.g., Wolff v. McDonnell, 418 U.S. 539 (1974); Vitek v. Jones, 445 U.S. 480 (1980); Washington v. Harper, 494 U.S. 210 (1990); Pratt v. Rowland, 65 F.3d 802, 807 (9th Cir. 1995). But when the entire sanction is confinement in disciplinary segregation for a period that does not exceed the remaining term of the prisoner's incarceration, it is difficult to see how after Sandin it can be made the basis of a suit complaining about a deprivation of liberty. Every state must have somewhere in its prison system single-person cells in which prisoners are sometimes confined not because they have misbehaved but simply because the prison has no other space, wishes to protect some prisoners from others, wishes to keep prisoners isolated from one another in order to minimize the risk of riots or other disturbances, wishes to prevent the spread of disease, and so forth. Almost 6 percent of the nation's prison inmates are in segregation, Criminal Justice Institute, Inc., Corrections Yearbook 22 (1997), and it appears that the great majority of these are not in disciplinary segregation (see Criminal Justice Institute, Inc., Corrections Yearbook: Adult Corrections 27 (1995), showing that in 1995 almost 5 percent of all prison inmates were in nondisciplinary segregation); so even a prisoner who had committed a white-collar crime and had been assigned to the lowest-security prison in the state's system might find himself in segregation for a nondisciplinary reason. Meriwether v. Faulkner, 821 F.2d 408, 414 (7th Cir. 1987). Under Sandin this possibility is probably enough to prevent him from complaining successfully of a deprivation of liberty if he is transferred into segregation for a disciplinary infraction.

This is a harsh result and perhaps the Court did not actually intend it; it is not discussed in the Sandin opinion; the Court remarked that Conner was an inmate of a maximum-security prison in which even nonsegregated inmates were kept in their cells between 12 and 16 hours a day, 515 U.S. at 475, 486 n. 8, which is considerably longer than in many prisons, especially minimum-security prisons. It is only when the transfer cases are brought to bear that the conditions in the prisoner's particular prison become irrelevant. And while the Court cited the transfer cases with approval in Sandin, see 515 U.S. at 478, it did not draw the logical inference and may not have intended to push its approach to its logical extreme — which would have made the conditions in Conner's prison irrelevant had it not been, in fact, the most restrictive prison in Hawaii's system. The question whether the comparison group includes other prisons did not have to be answered in Sandin, and is not discussed in either the majority opinion or any of the separate opinions. We consider ourselves bound by the Court's logic as well as its narrow holding, but we would welcome clarification of the issue by the Court.

A subsidiary issue on which authoritative guidance would also be most welcome is whether the comparison group can be confined to a single state. Indiana points out that it frequently swaps prisoners with other states pursuant to an interstate compact to which it is a party. Ind. Code sec. 11-12-8-2; see also Criminal Justice Institute, Inc., supra, at 46 (1997 ed.). It makes a good deal of sense to view the entire federal and state jail and prison system as a single system in order to balance the total prisoner load now well in excess of one million people. And this is how the prison authorities do view it. But this means that a prisoner in Indiana could well end up serving a portion of his sentence in another state, and that state might have even more restrictive conditions than Indiana. The logic of Sandin implies that the conditions of Wagner's disciplinary segregation are atypical only if no prison in the United States to which he might be transferred for nondisciplinary reasons is more restrictive. But we need not decide whether logic should be pressed so far. The likelihood that another state has more restrictive conditions than Indiana's Level V prisons is slight, given what would appear to be the small space between Level V as described to us by the state and the prohibitions of the Eighth Amendment.

Although it is extremely unlikely that Wagner can establish that the conditions of his disciplinary segregation are significantly more restrictive than conditions of confinement elsewhere within the Indiana prison system, the absence of any factual record in the district court on the issue makes it inappropriate for us to affirm out of hand. The judgment dismissing Wagner's suit is therefore vacated and the case remanded for further proceedings consistent with this opinion.

Vacated and Remanded, With Directions.