This is an appeal from a denial of a writ of habeas corpus. Petitioner argued in district court that the application to his case of a parole statute and related guidelines adopted subsequent to his sentencing is a violation of the ex post facto prohibition of the United States Constitution. The district court found that the new guidelines and statute could not be shown to create any additional disadvantage to the petitioner and denied his petition. 575 F.Supp. 284 (1983). We vacate and remand.
I.
Petitioner-Appellant, William J. Prater, is serving a life sentence imposed upon him in the Western District of Pennsylvania for the offense of conspiracy to injure, oppress or threaten a U.S. citizen. The conspiracy of which Mr. Prater was convicted resulted in the murder of United Mine Workers official Joseph Yablonski and two family members on December 31, 1969. This was a notorious murder and was widely deplored at the time.
*1232On February 3, 1982, two hearing examiners conducted Mr. Prater’s initial parole hearing at the United States Penitentiary at Terre Haute, Indiana, pursuant to the provisions of 28 C.F.R. § 2.13 (1981). After consideration of the relevant factors, the panel recommended parole effective August 8,1982. Subsequently, the application for parole was designated for original jurisdiction and referred to the National Commissioners pursuant to the provisions of 28 C.F.R. § 2.17 (1981).
On March 24,1982, the National Commissioners issued a notice of action which set a presumptive parole eligibility date of April 14, 1988. The reason stated by the commissioners for this action was that “release at this time would depreciate the severity of your offense behavior.” After an appeal to and hearing before the full commission in June of 1982, the commission affirmed the deferment of Mr. Prater’s parole to April of 1988.
Throughout Mr. Prater’s ten years of confinement he has maintained a clear institutional record of no disciplinary “write-ups” and has worked in prison industries; he had no prior convictions or history of drug dependence. He therefore received the highest possible “salient factor” score of ten (10). The special prosecutor investigating the Yablonski murders, Richard A. Sprague, and sentencing Judge Gerald J. Weber recommended clemency in view of Mr. Prater’s cooperation with the government in the prosecution of other conspirators.
On January 28, 1983, Prater filed a petition for writ of habeas corpus in the district court. One of the grounds for relief, and the only one raised on this appeal, is that the denial of parole for the reason that Prater’s release would depreciate the severity of his offense was in violation of the ex post facto clause of the United States Constitution.
The district court entered an order to show cause and the respondents responded on March 23, 1983. The response did not deny any of the factual allegations, but asserted that the respondents were entitled to dismissal of the petition as a matter of law. On May 16, 1983, the petitioner filed a motion for judgment on the pleadings noting the absence of a dispute as to any material fact and asserting that he was entitled to judgment as a matter of law.
The respondents filed a memorandum in response to the motion for judgment on the pleadings and supplemented their response by the filing of three additional exhibits relating to the Parole Commission’s action. Subsequently, the district court, ruling on the respondents’ motion to dismiss and petitioner’s motion for judgment on the pleadings and finding that no evidentiary hearing was necessary, granted the respondents’ motion and dismissed the petition. Petitioner Prater then took this appeal.
Petitioner’s argument on appeal is relatively straightforward. He pleaded guilty in 1973 to an offense committed in 1969 for which he was sentenced in June 1973. In 1969 and in 1973, the standard for parole release was governed by 18 U.S.C. § 4203 (1969) which required a determination that the prisoner would live at liberty without violating the law and that release was not “incompatible with the welfare of society.” 1 There were no formal regulatory *1233guidelines for the exercise of the Board’s discretion at the time of Mr. Prater’s offense or his plea of guilty.
In November 1973, the Parole Board adopted regulations which prescribed reasons for denial of parole and established guidelines for the length of time to be served based on the “salient factor” scores of individual characteristics and offense severity ratings. 38 Fed.Reg. 31942-45 (1973). In addition to the “institutional performance” and “public welfare” standards recognized in the then-applicable statute, the new regulations provided that the reasons for parole denial could include the factor that “[r]elease, in the opinion of the Commission, would depreciate the seriousness of the offense or promote disrespect for the law.” 28 C.F.R. § 2.13(d)(2) (1981).2 In 1976 Congress passed the Parole Commission and Reorganization Act of 1976, P.L. 94-233, 90 Stat. 219, codified at 18 U.S.C. § 4206(a)(1) (Supp.1982).3 Whereas the 1973 guidelines permitted the Parole Board to consider depreciation of the seriousness of the offense (thereby perhaps just codifying prior practice), Congress made that criterion a precondition of parole in the new act.
The “new” criterion contained in the 1973 guidelines and the 1976 statute relating to the depreciation of the seriousness of the offense and the promotion of disrespect for the law is a standard based on “retributive justice” (the relationship between time served and the nature of the offense) and “general deterrence” (incarceration as a means of promoting general respect for the law). See Heirens v. Mizell, 729 F.2d 449, 452 (7th Cir.), cert. denied, — U.S. —, 105 S.Ct. 147, 83 L.Ed.2d 85 (1984). The contention here, as in Heirens, is that the introduction of an *1234explicit general deterrence criterion into a parole scheme ostensibly guided solely by considerations of rehabilitation, risk of release to society and special deterrence (involving the characteristics of the prisoner himself and not the effect of his parole treatment on others) operates as a ex post facto law in violation of the Constitution. The argument is that this “new” criterion introduces a factor subsequent to the time of the commission of the crime and the time of the guilty plea and of sentencing, which makes it more difficult for the prisoner to win parole. We shall focus our analysis on the statutory change rather than on the introduction of the guidelines since a number of courts have held that such guidelines do not preclude or severely limit the exercise of discretion and therefore do not operate as ex post facto laws. Dufresne v. Baer, 744 F.2d 1543 (11th Cir. 1984); Warren v. United States Parole Comm’n, 659 F.2d 183 (D.C.Cir.1981), cert. denied, 455 U.S. 950, 102 S.Ct. 1454, 71 L.Ed.2d 665 (1982); Rifai v. United States Parole Comm’n, 586 F.2d 695 (9th Cir. 1978); Ruip v. United States, 555 F.2d 1331 (6th Cir.1977). This principle, which may, at least in a general way, have been approved by this circuit in Zeidman v. United States Parole Comm’n, 593 F.2d 806, 807 (7th Cir.1979), cannot in any event be applied to validate the 1976 statute which is before us.
II.
On a closely analogous problem this circuit has, in effect, marched to the top of the hill and then down again on the other side. Thus, in Welsh v. Mizell, 668 F.2d 328, 331 (7th Cir.), cert. denied, 459 U.S. 923, 103 S.Ct. 235, 74 L.Ed.2d 186 (1982), we held that the explicit introduction of similar retributive general deterrent considerations into the Illinois Parole statute operated as an ex post facto law and could not be applied to prisoners whose crimes were committed before the adoption of the statutory change. In Heirens, supra, however, we overruled Mizell because we determined from an examination of parole practice in Illinois that general deterrence had in fact been a valid consideration by the Parole Board even before the statutory revision which made this consideration one for explicit examination. In other words, we determined that, with respect to the Illinois statute, the adoption of a “depreciation of the seriousness of the offense” standard merely made explicit what had formerly been implicit in the administration of the parole statute and therefore there had been no violation of the ex post facto prohibition.
The facts of Heirens are somewhat like the facts in this case. William Heirens had been sentenced in 1946, in Illinois courts, to life imprisonment for each of three murders, and had been sentenced for other crimes to lesser terms, to run consecutively to the three life terms, which were themselves to run consecutively. In 1981 Heir-ens filed a petition for a writ of habeas corpus in federal court; he had been denied parole for the reason that release would depreciate the seriousness of his crime, and he claimed that the imposition of that criterion, adopted by the legislature in 1972, was in his case a violation of the ex post facto clause of the Constitution. While his petition was before the district court, we decided Welsh, saying that the “depreciation of the seriousness” standard was a marked departure from previous practice, and that its application to Welsh, who had been sentenced before the Illinois legislature had promulgated that standard, was indeed an ex post facto violation. The magistrate who had been assigned Heirens’ petition ordered the Parole Board to give Heirens a new hearing, under Welsh, applying standards in effect at the time of Heirens’ sentencing. The Parole Board made several attempts to justify the denial of parole but could not convince the magistrate that parole was being denied for reasons other than the retroactively inapplicable general deterrence rationale. The magistrate therefore ordered Heirens’ release, and the Parole Board appealed.
In overruling Welsh, this aourt began by looking closely at existing practice before the passage of the Illinois statute. It *1235found that although rehabilitative progress was considered by the Parole Board under the older practice, it was by no means the only consideration.
[TJhe Parole Board was effectively given unlimited discretion in deciding whether or not to parole a particular prisoner with the single exception that the Board include in its consideration the prisoner’s conduct record while confined. We cannot conclude, as the court did in Welsh, that the statute’s requirement that the Board consider an applicant’s prison record eliminated all other pertinent criteria, including general deterrence and retributive justice, from its lawful consideration, when the Parole Board’s discretion was otherwise unlimited.
729 F.2d at 460 (emphasis in original). We quoted from People v. Nowak, 387 Ill. 11, 55 N.E.2d 63, cert. denied, 323 U.S. 745, 65 S.Ct. 67, 89 L.Ed. 597 (1944), in which the Illinois Supreme Court had said:
[Benefits under the provisions of the Parole Act] may be granted as a matter of grace and not as a legal right. Such acts of leniency, whether by pardon or parole, are administered by the executive branch of the government in the interests of society and the discipline, education and reformation of the one convicted.
729 F.2d at 462 (quoting Nowak, 55 N.E.2d at 65) (emphasis in Heirens). The Illinois court had thus made clear that in addition to education and reformation, the interests of society and discipline were to be taken into account in awarding parole.
We also pointed in Heirens to an article by Theodore Fields, a former Illinois Parole Board chairman, in which Mr. Fields discussed the new parole legislation. We quoted:
The duties of the Parole Board are set forth by the Illinois statutes. The new statute which became law in January, 1973, was drafted by progressive experts in the corrections field. What did they provide as to the factors that the Board must take into consideration when making parole decisions? The statute sets forth that “the Board shall not parole a person eligible for parole if it determines that:
“(1) There is a substantial risk that he will not conform to the reasonable conditions of parole; or
“(2) his release at that time would depreciate the seriousness of his offense or promote disrespect for the law; or
“(3) his release would have a substantially adverse effect on institutional discipline.”
This statute sets forth the general criteria for denying parole that have been followed by the Board for some time.
729 F.2d at 460 (quoting Fields, Illinois Parole and Pardon Board Adult Parole Decisions, 62 Ill.B.J. 20, 20-21 (1973)) (emphasis in Heirens).
We concluded:
[The Illinois legislature] merely codified the Board’s prior practice and procedure, that is, it simply explicitly articulated the Parole Board’s broad range of discretion which had always existed. Since the Parole Board considered both general deterrence and retributive justice prior to 1973, the application of [the depreciation of the seriousness criterion] to inmates who committed their crimes before 1973 does not violate the ex post facto prohibition of the United States Constitution. [That] criterion is not disadvantageous to an offender who committed his crime before January 1, 1973, as that criterion merely includes factors which were considered in making parole decisions prior to that date.
729 F.2d at 463.
III.
A.
“[¡TJwo critical elements must be present for a criminal or penal law to be ex post facto: it must be retrospective, that is, it must apply to events before its enactment, and it must disadvantage the offender affected by it.” Weaver v. Graham, 450 U.S. 24, 29, 101 S.Ct. 960, 964, 67 *1236L.Ed.2d 17 (1981) (footnotes omitted).4 In this case, unlike Weaver, the question of disadvantage cannot be decided as a matter of law; as in Heirens, disadvantage here is a factual question involving the actual practice of the Parole Board before and after the statutory change. Weaver involved a change in the Florida “gain-time” statute which prescribed the deductions from their sentences which prisoners could win as gain-time for good conduct. The Court in that case looked no farther than the statute, on the face of which the disadvantage was clear and demonstrably significant. Thus, the statute provided that the number of days per month off for good conduct was to be reduced from five days, ten days and fifteen days per month, to three days, six days and nine days per month, respectively, depending on the year of sentence involved. The disadvantage to the offender of the changed formula for calculating “gain-time credits” was clear as a simple matter of arithmetic: three days is less than five days, six days is less than ten days and nine days is less than fifteen days. In Weaver, the Supreme Court also noted that the statutory change would result in significantly extending the time Weaver was required to remain in prison by over two years, or approximately fourteen percent of his original fifteen-year sentence.
In the case before us, as in Heirens, disadvantage is a factual question involving the actual practice of the Parole Board before and after the statutory change. The district court evidently considered the question to be a matter of law, and found no ex post facto violation. In a very brief paragraph the district judge drew the conclusion that the petitioner could not have been disadvantaged by the change in the law:
It is at this point that petitioner’s ex post facto argument fails, because under the law as it existed in 1969, at the time of the petitioner’s offense, it would have been possible for the Board of Parole (the predecessor to the Parole Commission) to deny parole for the reason that release would depreciate the seriousness of the offense. Thus petitioner could not have been disadvantaged by his denial of parole for that reason in 1982.
But speculation that the Board might have acted in the same way under the old law does not settle the question of disadvantage, where the question is being decided as a matter of law. If we have nothing to go on but the assumption that parole could have been denied under the old law if it would depreciate the seriousness of the offense, and the requirement that parole must be denied under the new law if it would depreciate that seriousness, the prisoner could at least argue persuasively that he had been disadvantaged. Under the older statute the board was apparently entitled, in its discretion, to balance depreciation of seriousness against other factors, especially in borderline cases of such depreciation. Under the newer statute, no one could be released whose parole would depreciate the seriousness of the offense, whatever else was in his favor. The probability of release of someone whose release would thus depreciate his crime would tend to decline and this change was arguably to his disadvantage.
We feel, therefore, that the court below was incorrect in deciding as a matter of law that Prater could not have been disadvantaged by the change in the statute. It seems to us simply impossible to draw this conclusion out of the statutory language alone.
In Heirens, however, we held that where such a change in discretion involves the parole laws, the courts must look to the actual practice of the board before and after the change, and must not try to decide the question of disadvantage as a matter of law. The practical approach relied on in Heirens seems to us the right one, one grounded in the very nature of the *1237parole laws and of the considerations on which parole boards necessarily rely.5
Certainly it is true, as the Supreme Court said in Weaver, that a change which required that prisoners receive fewer days for good time would lower each prisoner’s expectation of early release; in that case the discretion of the parole board is not involved. Even though a prisoner might get out just as early by qualifying for extra days of good time, the Court did not countenance such speculation. 450 U.S. at 85-36, 101 S.Ct. at 967-968. The probability of early release was reduced, and that sufficed to violate the ex post facto laws when applied retroactively. Similarly, a change from a discretionary maximum sentence to a mandatory sentence of the same length would change expectations for the worse; under the earlier law each prisoner had some positive expectation of getting less than the maximum sentence, but under the later law he has no such expectation. Lindsey v. Washington, 301 U.S. 397, 401, 57 S.Ct. 797, 799, 81 L.Ed. 1182 (1937). Even though he might have gotten the maximum sentence under the old law, the change in expectations is clearly to his disadvantage, and would violate the ex post facto law. Weaver, 450 U.S. at 32 n. 17, 101 S.Ct. at 966 n. 17 (citing Lindsey v. Washington, supra).
But when we come to the necessarily elusive factors which must enter into the parole board’s determination about release, such conclusions are inappropriate. Although on paper it can be made to look as though any law restricting the grounds on whiéh a prisoner could be paroled worked to his disadvantage, that is not really the case. Consider a law which, unlike prior law, required the board to deny release unless it seemed to the board that release was for the good of all. Such a law would probably have no effect whatever on the practice of the board; it would merely tell the board not to release someone unless they thought it was the right thing to do, something we may assume the board has taken for granted all along. And yet it is beyond question that one area of the board’s discretion — the discretion to parole even when it is not for the good of all— would have been cut off; and if these things had to be decided in the abstract, it would have to be said that the prisoner had been accordingly disadvantaged.
In our view, such a change would not violate the ex post facto clause, because no real disadvantage would have been suffered. Hence the question here is whether the actual change that was made by the 1976 law is like the “for-the-good-of-all” change, or whether in fact it disadvantages the prisoner. And that is a question of fact, as we made clear in Heirens, to be determined by an investigation of the practice of the board before and after the change. We would note that the “good-of-all” language in our example is not radically different from the sort of expansive language the Illinois Parole Board and the United States Parole Commission have been obliged to deal with.
B.
Since the court below considered this to be a question of law, it did not take evidence on the practice of the parole board. We remand for a reconsideration in the light of what we have said here and in Heirens. While Heirens involved a different statute and a different board, the statutes are nearly identical and the forces at work to produce the change were quite obviously similar. Before we will conclude that the change in the federal parole law has worked to Prater’s disadvantage, therefore, we must be convinced that the change in the statute worked a change in the actual practice of the federal parole authorities.
*1238The ultímate burden of showing that a statutory change would result in clear and significant disadvantage to the parolee must be upon the party asserting unconstitutionality. Adams v. United States ex rel. McCann, 317 U.S. 269, 281, 63 S.Ct. 236, 242, 87 L.Ed. 268 (1942); United States ex rel. Williams v. DeRobertis, 715 F.2d 1174, 1178 (7th Cir.1983). In this case, this is the petitioner. In Heir-ens, we relied on scholarly discussions and prior court cases to determine what the actual practice of the board had been, and to what extent it changed. Unfortunately, no such clear evidence is available to us here, and we must rely on the parties to establish the practice. And, of course, the petitioner must be given an opportunity to present evidence if he has any.
Our own perusal of the legislative history and the commentary unearth nothing which conflicts with the Heirens’ assumption that parole boards before 1970 did in fact consider depreciation of the seriousness of the offense in making their determinations. While the legislative history of the United States parole statute and the commentary relating to it are not conclusive on the point, there is indeed some significant commentary to the effect that the United States Board of Parole considered general deterrence, retributive justice and exemplary considerations before the adoption of the 1973 guidelines or the 1976 statute:
[T]he former chairman of the United States Board of Parole [George Reed] has provided an example of board maintenance of general deterrence aims:
You take a bank president who embezzles half a million dollars and he gets a substantial sentence — the judgment of the court is (A) he is going to observe the rulings of the institution, no question about that — he is not going to be a problem in the institution — and (B) probably if released he would not be the president of the bank, he would not have the opportunity to embezzle money. But there is a matter of public interest and the public welfare that there be some deterrent from people taking half a million dollars out of your bank or mine. In making decisions premised on this approach, the Board is really considering the “weightiness of considerations of retribution, moral reprobation, community reassurance or deterrence.”
Kastenmeier and Egglett, Parole Release Decisionmaking: Rehabilitation, Expertise, and the Demise of Mythology, 22 Am.U.L.Rev. 477, 508 (1973) (footnotes omitted).
And the same authors note that the former (pre-1973) statutory mandate for the Board was that it consider the “welfare of society.”
“Welfare,” obviously a vague term, lends itself to definitions ranging from public safety to public adverse reaction. Consequently, notwithstanding the possibility of a man’s readiness for release in terms of his own personal makeup and even gauged by the measure of public safety, concern about public opinion [presumably about dangerousness] may deter an affirmative decision to release.
Id. at 516 (emphasis supplied; footnotes omitted). Thus, it appears that, although explicit recognition in administrative guidelines and in the applicable statute is relatively recent, the United States Board of Parole may have for a long time if not always — recognized and deferred to considerations of general deterrence, retribution and the like. See Warren v. United States Parole Comm’n, 659 F.2d 183, 193 (D.C. Cir.1981) (“[T]he guidelines embody what may well have been the Board’s practice anyway.”). At least, we know of no basis in the legislative history or commentary whereby the impact of new guidelines and statutory provisions in the federal arena must, or should, be considered essentially different from the like circumstances with respect to parole in the state of Illinois, which we examined in Heirens. Nevertheless we are unable to find dispositive evidence that the statutory changes had no real effect on the practice of the parole authorities.
*1239The petitioner’s burden, of course, is not to show that the change worked to his own actual disadvantage, but to show that the operation of the later statute is significantly more onerous in its effect than the operation of the former statute. Weaver v. Graham, 450 U.S. at 33, 101 S.Ct. at 966 (“the inquiry looks to the challenged provision, and not to any special circumstances that may mitigate its effect on the particular individual”); Dobbert v. Florida, 432 U.S. 282, 292, 97 S.Ct. 2290, 2298, 53 L.Ed.2d 344 (1977) (fact that a change was “on the whole ameliorative” was an “independent basis” for conclusion that there was no violation of ex post facto clauses); Lindsey v. Washington, 301 U.S. 397, 401, 57 S.Ct. 797, 799, 81 L.Ed. 1182 (1937) (“It is true that petitioners might have been sentenced to fifteen years under the old statute. But the ex post facto clause looks to the standard of punishment prescribed by a statute, rather than to the sentence actually imposed.”) That means that he is not called on to prove to our satisfaction that he personally was injured, any more than a gambler who has been cheated has to show that he would have won in the absence of cheating in order to get a new deal.
IV.
A.
The policy of the ex post facto clause is to insure that persons have fair notice of potential criminal punishment and will be able to rely for their conduct on the criminal law as it exists at the time of their acts. Reliability and regularity are the goals. See Weaver v. Graham, 450 U.S. 24, 28-29, 101 S.Ct. 960, 963-964, 67 L.Ed.2d 17 (1981); Warren v. United States Parole Comm’n., 659 F.2d 183, 188 (D.C.Cir.1981); L. Tribe, American Constitutional Law, 474, 484 (1978). It seems implausible that the perpetrator of a grave and notorious crime could operate under a serious misapprehension that a parole board, armed with broad discretion, would fail to consider the public ramifications of his premature release from prison — whatever the formal state of the applicable statutes or guidelines.
Nevertheless, it may in fact be so; it may be that the 1976 statute changed parole practice and that Prater has been disadvantaged to some extent by the change in the law. We note that in his pleadings before the district court Prater raised factual issues concerning actual disadvantage to himself. He did not, however, address the question of parole practice. The trial judge, working under an erroneous view of the law, did not address the factual issue but ruled against Prater on the basis of his reading of the statute itself. Under these circumstances we must remand to allow consideration of the question of practice. If Prater can demonstrate, to the satisfaction of the district court, that his expectations of release at a certain time have been clearly and significantly reduced, then he should be permitted to invoke the ex post facto clause.
B.
The dissent suggests that we are inviting an increase in litigation by concluding that the change in practice involved in this case might implicate the ex post facto clause. That may or may not be a relevant consideration in approaching a constitutional provision. Whether it is or not, we foresee no such increase in litigation. As in Heirens, once the question of a violation has been decided in this case, future litigation should be foreclosed.
Moreover, if the dissent’s primary concern is with litigation, it has chosen an odd way out of the difficulty it perceives. Whereas we have suggested that a determination about practice must be made that would settle the matter once and for all, the dissent would avoid remanding by finding the later statute “more liberal” in creating an entitlement to parole (providing the conditions of the statute are met). In the first place, finding such an entitlement in the language of the act itself suggests endless litigation. In the second place, unless there is a kind of ratchet that guarantees that all future statutes will be even *1240more liberal, the dissent’s approach only defers the ex post facto problem; for should some future statute not contain the same liberal language, would prisoners not be right, under the dissent’s reasoning, to argue that that new statute should not apply retroactively to them? We note, in fact, that under the new Crime Control Act of 1984, parole release is eliminated and all release dates for prisoners incarcerated under the 1976 statute will be set at the end of five years. Crime Control Act § 235(b)(1)(A), (b)(2)-(4), 98 Stat. 2032-33 (1984). Will prisoners whose release determination is thus accelerated have an ex post facto claim because of the “liberality” of the 1976 statute? Will it have to be determined on a case by case basis whether the early determination violated their entitlement?
But even should the dissent be right, that the later statute creates an entitlement to parole, the inquiry cannot end there. The statute also creates a class for whom parole was formerly, under the law, within the discretion of the parole board and for whom parole is now statutorily impossible — namely, the class of prisoners whose release would depreciate the seriousness of their crime. See H.R. Conf.Rep. No. 94-838, 94th Cong.2d Sess. 26, reprinted in [1976] U.S.Code Cong. & Ad.News 335, 351, 358 (“the release on parole of only those who meet the criteria of this act”). On the face of it, this class has been disadvantaged. But Heirens tells us to look beyond the words of the statute and to decide disadvantage as a question of fact. Since we do not have at our disposal the evidence that was apparently before the Heirens court, and since we are reluctant to take the dissent’s advice and base a different outcome on a passage from a law review article, we are compelled to remand.6
The decision of the district court is vacated, and the cause is remanded for further proceedings not inconsistent with this opinion.
. § 4203. Application and release; terms and conditions
(a) If it appears to the Board of Parole from a report by the proper institutional officers or upon application by a prisoner eligible for release on parole, that there is a reasonable probability that such prisoner will live and remain at liberty without violating the laws, and if in the opinion of the Board such release is not incompatible with the welfare of society, the Board may in its discretion authorize the release of such prisoner on parole.
Such parolee shall be allowed in the discretion of the Board, to return to his home, or to go elsewhere, upon such terms and conditions, including personal reports from such parole persons, as the Board shall prescribe, and to remain, while on parole, in the legal custody and under the control of the Attorney General, until the expiration of the maximum term or terms for which he was sentenced. Each order of parole shall fix the limits of the parolee's residence which may be changed in the discretion of the Board.
*1233(b) The parole of any prisoner sentenced before June 29, 1932, shall be for the remainder of the term or terms specified in his sentence, less good time allowances provided by law.
18 U.S.C. § 4203 (1969).
. § 2.13(d).
... [Rjeasons for parole denial may include the following, with further specification as appropriate:
(1) The prisoner has not substantially observed the rules of the institution or institutions in which confined;
(2) Release, in the opinion of the Commission, would depreciate the seriousness of the offense or promote disrespect for the law; or
(3) Release, in the opinion of the Commission, would jeopardize the public welfare. In lieu of, or in combination with, the above reasons the prisoner shall be furnished with a guidelines evaluation statement____
28 C.F.R. § 2.13(d).
. § 4206. Parole determination criteria
(a) If an eligible prisoner has substantially observed the rules of the institution or institutions to which he has been confined, and if the Commission, upon consideration of the nature and circumstances of the offense and the history and characteristics of the prisoner, determines:
(1) that release would not depreciate the seriousness of his offense or promote disrespect for the law; and
(2) that release would not jeopardize the public welfare; subject to tlib provisions of subsections (b) and (c) of this section, and pursuant to guidelines promulgated by the Commission pursuant to section 4203(a)(1), such prisoner shall be released.
Pub.L. 94-233, § 2, 90 Stat. 223 (1976).
Prior to the passage of this statute, 28 C.F.R. § 2.18, "Granting of Parole,” had read as follows:
The granting of parole rests in the discretion of the Board of Parole. The Board may parole a prisoner who is otherwise eligible if
(a) in the opinion of the Board such release is not incompatible with the welfare of society;
(b) he has observed substantially the rules of the institution in which he is confined; and
(c) there is a reasonable probability that he will live and remain at liberty without violating the laws.
Subsequent to the passage of the statute, that section of the guidelines was changed to read as follows:
The granting of parole to an eligible prisoner rests in the discretion of the United States Parole Commission. As prerequisites to a grant of parole, the Commission must determine that the prisoner has substantially observed the rules of the institution or institutions in which he has been confined; and upon consideration of the nature and circumstances of the offense and the history and characteristics of the prisoner, must determine that release would not depreciate the seriousness of his offense or promote disrespect for the law, and that release would not jeopardize the public welfare (i.e., that there is a reasonable probability that, if released, the prisoner would live and remain at parole).
. The Weaver Court noted that merely procedural changes in law did not violate the ex post facto provision. See Dobbert v. Florida, 432 U.S. 282, 293, 97 S.Ct. 2290, 2298, 53 L.Ed.2d 344 (1977).
. In United States ex rel. Forman v. McCall, 709 F.2d 852 (3d Cir.1983), the Third Circuit found the following issue to be a question of fact, rather than law: whether particular parole guidelines “have the effect of ‘laws’ on the Commission's decisionmaking or whether they constitute mere guideposts facilitating the exercise of the Commission’s discretion ..." Id. at 853. As we have seen, other circuits have held as a matter of law that parole guidelines are not laws for the purposes of the ex post facto clause.
. The dissent says our result is not compelled by Heirens, and points to a footnote in which Heir-ens says that the lower court decided this case on grounds "in accord with the present case [i.e., Heirens ].” That is indeed one way of reading the lower court opinion. Unfortunately, however, the judge below, reaching his decision before Heirens was handed down, failed to make the appropriate findings. Even if he did accept the legal premise Heirens is based on, he did not conduct the same painstaking inquiry into the Commission's procedures, and the case must be remanded for findings of fact. We think, however, that the district judge in this case decided against Prater as a matter of law.