ARNOLD, Circuit Judge.
The appellant, Billy Woodard, has been tried by jury, found guilty of capital felony murder, and sentenced to death. See 261 Ark. 895, 553 S.W.2d 259. Woodard was tried by a so-called “death-qualified jury,” from which all persons who refused to consider the death penalty had been excluded. In
Grigsby v. Mabry,
758 F.2d 226 (8th Cir.1985) (en banc), we held that such juries are not fairly representative of the community and are not impartial on the question of guilt or innocence. In Woodard’s case we must face the issue whether
Grigsby
is to be applied “retroactively,” that is, whether Woodard, whose jury was selected before this Court decided
Grigsby,
is entitled to the benefit of the rule announced in that case. We hold that
Grigsby
does apply to Woodard’s case and that no procedural default bars Woodard from raising the point in this federal habeas corpus proceeding. We therefore reverse the judgment of the District Court, which dismissed Woodard’s petition, and remand this cause to it with directions to grant the writ unless the State of Arkansas, within such reasonable time as the District Court may fix, commences proceedings to retry Woodard before an impartial jury.
I.
“As a rule, judicial decisions apply ‘retroactively.’
Robinson v. Neil,
409 U.S. 505, 507-08 [93 S.Ct. 876, 877-78, 35 L.Ed.2d 29] (1973).”
Solem v. Stumes,
— U.S.—, 104 S.Ct. 1338, 1341, 79 L.Ed.2d 579 (1984).
Accord, United States v. Sager,
743 F.2d 1261, 1265 (8th Cir.1984). But ever since
Linkletter v. Walker,
381 U.S. 618, 85 S.Ct. 1731, 14 L.Ed.2d 601 (1965), the Supreme Court has held that “the interest of justice” and “the exigencies of the situation” may argue against imposing retroactively a new rule of law enlarging the rights of criminal defendants.
Id.
at 628, 85 S.Ct. at 1737. The Court’s most recent full-dress treatment of the subject is its opinion in
Solem v. Stumes, supra.
There, as to cases, like the one before us, not involving the Fourth Amendment, the Court reiterated that the issue of retroac-tivity depends on three factors: (a) the purpose to be served by the new rule; (b) the extent to which law-enforcement authorities have relied on an older rule of law; and (c) the effect on the administration of justice of a retroactive application of the new rule. 104 S.Ct. at 1341.
The balance among these factors varies with the context in which they are
applied. In each case, the purpose of the new rule of law, the degree of justifiable surprise that it produces in the minds of the authorities, and the effect of its retroactive application on the administration of justice, must all be examined. Here, we are fortunate in having the guidance of a Supreme Court opinion that is virtually on all fours with the present case. That opinion is
Witherspoon v. Illinois,
391 U.S. 510, 88 S.Ct. 1770, 20 L.Ed.2d 776 (1968), itself, the case that originated the whole line of recent analysis of the death-qualified jury.
The petitioner in
Witherspoon
made two main arguments: (1) that the exclusion for cause from his jury of persons with conscientious scruples against the death penalty made that jury less than impartial on the question of guilt or innocence; and (2) that the same jury was, in any event, not impartial on the issue of punishment. The Court rejected the first argument, not for any analytical fault or intrinsic lack of merit, but for failure of proof on the particular record before it. It accepted the second argument. Thus, Witherspoon’s sentence, but not his conviction, was overturned.
For present purposes, the important thing about
Witherspoon
is that the Court specifically declined to make its holding on the second issue prospective only. A brief
amicus curiae
filed by 27 States, pointing to the statement in
Logan v. United States,
144 U.S. 263, 298, 12 S.Ct. 617, 628, 36 L.Ed. 429 (1892), that it was permissible to exclude from capital juries those who had conscientious scruples against the death penalty, asked the Court (assuming its determination to depart from
Logan)
not to make any new rule retroactive. The Court weighed the three criteria given above and squarely rejected this request:
[T]he jury-selection standards employed [at Witherspoon’s trial] necessarily undermined “the very integrity of the ... process” that decided the petitioner’s fate, and we have concluded that neither the reliance of law enforcement officials, nor the impact of a retroactive holding on the administration of justice, warrants a decision against the fully retroactive application of the holding we announce today.
Witherspoon,
391 U.S. at 523 n. 22, 88 S.Ct. at 1777-78 n. 22 (citations omitted).
The new rule announced in
Grigsby,
of course, is not the same new rule announced in
Witherspoon,
though it was foreshadowed by the
Witherspoon
opinion,
id.
at 520 n. 18, 88 S.Ct. at 1776 n. 18. The
Witherspoon
rule is that juries from which jurors with any conscientious scruple against the death penalty have been excluded are not impartial on the issue of life or death. The
Grigsby
rule is that juries from which jurors who refuse to consider the death penalty have been excluded are not impartial on the issue of guilt or innocence.
We see no relevant difference between these two rules for purposes of the Supreme Court’s three retroactivity criteria. Like
Witherspoon,
the purpose of
Grigsby
is to make jury verdicts more reliable, to purge them of partiality, to make it less likely that the innocent will be convicted (or, as in
Witherspoon,
that a defendant unworthy of death will receive that extreme penalty). This purpose goes to the heart of the truth-finding function of a trial by jury. We doubt that law-enforcement officials are markedly more surprised by
Grigsby,
which was foreshadowed in
Witherspoon,
than they were by
Wither-spoon’s
rejection of
Logan.
Nor do we believe that a significantly greater number of convictions will be endangered by
Grigsby,
than sentences of death were by
Witherspoon.
The Supreme Court has weighed the three retroactivity factors for us in an indistinguishable context, and the balance struck by the Court clearly indicates that
Grigsby
should be given fully retroactive effect.
Probably the two cases that come closest to supporting a ruling against retroactivity here are
Daniel v. Louisiana,
420 U.S. 31, 95 S.Ct. 704, 42 L.Ed.2d 790 (1975) (per curiam), and
DeStefano v. Woods,
392 U.S. 631, 88 S.Ct. 2093, 20 L.Ed.2d 1308 (1968) (per curiam).
Daniel
holds that
Taylor v.
Louisiana,
419 U.S. 522, 95 S.Ct. 692, 42 L.Ed.2d 690 (1975), which invalidated a conviction returned by a jury from which women had been systematically excluded, is not to be applied retroactively. And
DeStefa-no
reaches the same conclusion as to the rules announced in
Duncan v. Louisiana,
391 U.S. 145, 88 S.Ct. 1444, 20 L.Ed.2d 491 (1968) (States may not deny jury trial in serious criminal cases), and
Bloom v. Illinois,
391 U.S. 194, 88 S.Ct. 1477, 20 L.Ed.2d 522 (1968) (right to trial by jury extends to serious criminal contempts). As to
DeStefano,
it was handed down only two weeks after
Witherspoon,
which it does not cite, and it is not to be supposed that the Court so soon decided to change its mind.
DeStefano
can hardly be read, therefore, to cast any doubt on the
Wither-spoon
retroactivity passage quoted above.
But more fundamentally, the rules of law announced in
Taylor, Duncan,
and
Bloom,
respectively, are all less directly related to the need for impartiality in the search for truth than are either
Witherspoon
or
Grigsby. Taylor
is based on the abstract ideal that juries should be representative of the community,. not on any theory that juries composed of men only, or with few women members, are less likely to reach accurate verdicts.
Duncan
is simply an application of the Sixth Amendment’s command to the States. It does not depend on any notion that juries are more likely to be impartial than judges. The rule in
Bloom
is perhaps closer for present purposes to that in
Grigsby,
because judges who are themselves the objects of contemptuous behavior may be less impartial in judging that behavior than juries would be, but that was only “[o]ne ground” underlying the holding in
Bloom. DeStefano,
392 U.S. at 634, 88 S.Ct. at 2095.
Grigsby,
by contrast, holds not only that juries shorn of
Witherspoon
excludables are not a fair cross section of the community, but also that they are prone to convict and not impartial. “[A] petit jury without WEs is ‘conviction prone’ and, therefore, not an impartial or cross-representative jury.”
Grigsby,
at 237. This is necessarily so, because the group (WEs) whose exclusion prevents the jury from being a fair cross section is defined on the basis of an attitude towards criminal justice. It is the resulting attitude of those who remain on the jury that makes them more likely to be conviction-prone and therefore less than impartial on the issue of guilt or innocence.
The purpose of
Grigsby,
in short, is “to overcome an aspect of the criminal trial that substantially impairs its truth-finding function and so raises . serious questions about the accuracy of guilty verdicts in past trials ____”
Williams v. United States,
401 U.S. 646, 653, 91 S.Ct. 1148, 1152, 28 L.Ed.2d 388 (1971) (plurality opinion),
approved in Hankerson v. North Carolina,
432 U.S. 233, 241, 97 S.Ct. 2339, 2344, 53 L.Ed.2d 306 (1977). In such cases, “[njeither good-faith reliance by state or federal authorities on prior constitutional law or accepted practice, nor severe impact on the administration of justice has sufficed to require prospective application ....” 401 U.S. at 653, 91 S.Ct. at 1152. See also
Solem v. Stumes, supra,
104 S.Ct. at 1342 (“Complete retroactive effect is most appropriate where a new constitutional principle is designed to enhance the accuracy of criminal trials.”);
United States v. Johnson,
457 U.S. 537, 544, 102 S.Ct. 2579, 2584, 73 L.Ed.2d 202 (1982);
Brown v. Louisiana,
447 U.S. 323, 328, 100 S.Ct. 2214, 2219, 65 L.Ed.2d 159 (1980) (giving retroactive effect, at least as to cases pending on direct review, to
Burch v. Louisiana,
441 U.S. 130, 99 S.Ct. 1623, 60 L.Ed.2d 96 (1979), invalidating convictions by nonuna-nimous six-person juries).
In short, fully retroactive application of
Grigsby
is supported both by the balance among the three retroactivity criteria struck in
Witherspoon
itself, and by more recent formulations of Supreme Court retroactivity doctrine. We hold that
Grigs-by
applies to Woodard’s case. The State argues, however, that Woodard has lost his right to urge the point on federal habeas corpus because he failed to preserve the argument properly in the state courts. To that question we now turn our attention.
II.
The District Court held that Woodard failed to raise his
Grigsby
claim properly in the state courts; that the Supreme Court of Arkansas, when it denied Woodard’s petition for post-conviction relief, rejected the
Grigsby
claim without reaching its merits because it had not been properly raised; that Woodard had not shown “cause” to justify his procedural default; and that therefore federal habeas review of this claim is barred. 567 F.Supp. at 1568-73. The District Court’s treatment of these issues is thoughtful and balanced, but after careful review of state law and the relevant opinions of the Supreme Court of Arkansas, we come to a different conclusion. In our view, the more likely reading of the opinion of the Supreme Court of Arkansas is that it considered and rejected Woodard's
Grigsby
argument on its merits. Therefore, federal habeas review is not barred.
See
Ulster County Court v. Allen,
442 U.S. 140, 148-49, 99 S.Ct. 2213, 2220, 60 L.Ed.2d 777 (1979);
Euell v. Wyr-ick,
675 F.2d 1007, 1008-09 (8th Cir.1982).
There is no doubt that Woodard, or his then counsel, is open to a charge of procedural default. He did not raise the
Grigsby
point either at trial or on direct appeal to the Supreme Court of Arkansas. It was first raised in his state-court post-conviction proceeding, a petition filed with the Supreme Court of Arkansas for leave to proceed under Ark.R.Crim.P. 37. This Rule “affords a remedy when the sentence was imposed in violation of the Constitution or laws of the United States or of this state or ‘is otherwise subject to collateral attack.’ ”
Hulsey v. State,
268 Ark. 312, 313, 595 S.W.2d 934, 935 (1980) (quoting Ark.R.Crim.P. 37.1). The
Grigsby
claim is based on the allegation that the Constitution of the United States is violated, so it comes within the literal language of the Rule. It is nevertheless well established that arguments that could have been, but were not, raised at trial and on direct appeal, even arguments based on the federal Constitution, are not reachable under Rule 37 unless they are “so fundamental as to render the sentence and the judgment void and open to collateral attack.”
Hulsey,
268 Ark. at 315, 595 S.W.2d at 936.
With these principles of state procedural law in'mind, we turn to the opinion of the Arkansas Supreme Court denying Woodard’s Rule 37 petition.
Woodard v. State,
273 Ark. 235, 617 S.W.2d 861,
cert. denied,
454 U.S. 1068,102 S.Ct. 618, 70 L.Ed.2d 603 (1981). That opinion is of course the first place to look to determine why the state court rejected Woodard’s
Grigsby
claim. “[I]f neither the state legislature [
] nor the state courts indicate that a federal constitutional claim is barred by some state procedural rule, a federal court implies no disrespect for the State by entertaining the claim.”
Ulster County,
442 U.S. at 154, 99 S.Ct. at 2223 (footnote omitted).
The Supreme Court’s opinion does not on its face clearly answer this question one way or the other. It does not expressly mention the
Grigsby
point at all. A passage rejecting a
Witherspoon
claim could be read to include
Grigsby,
because it refers to
“Witherspoon ...
and its progeny,” 273 Ark. at 239, 617 S.W.2d at 863, and
Grigsby
might be considered one of the “progeny” of
Witherspoon.
But probably it should not be so understood, because the concluding sentence in the paragraph in question, which we set out in full in the margin,
refers to Woodard’s sentence, not
his conviction, and
Grigsby,
unlike
Wither-spoon,
goes to the validity of the conviction itself, not merely of the sentence. At another point in its opinion the Court says, “Petitioner raises the issue of the death qualified jury,”
Id.
at 238, 617 S.W.2d at 863, but this sentence is not further explained at that point, and it could refer to
Witherspoon
alone.
Then, at the end of its opinion, the Court admonishes petitioner as follows:
It appears that the petitioner does not understand the purpose of Criminal Procedure Rule No. 37. It is not intended to provide a review of mere error in the conduct of the trial or to serve as a substitute for appeal.
Clark v. State,
255 Ark. 13, 498 S.W.2d 657 (1973). It also seems that petitioner feels that any error is subject to collateral attack. Unless an allegation raises issues so fundamental as to render the judgment void, such issues cannot be raised by the use of Rule 37.
Orman v. Bishop,
245 Ark. 887, 435 S.W.2d 440 (1968);
Moore v. Illinois,
408 U.S. 786 [92 S.Ct. 2562, 33 L.Ed.2d 706] (1972). Issues not raised by appellant in his original appeal are considered waived unless they are so fundamental as to render the judgment void.
Wicks v. State,
270 Ark. 781, 606 S.W.2d 366 (1980).
Id.
at 240, 617 S.W.2d at 864. It is clear enough that the Court is rejecting some of Woodard’s arguments on procedural grounds. But Woodard made a good many arguments not expressly mentioned in the Court’s opinion, and this paragraph does not tell us either that the
Grigsby
point is being rejected because of procedural default, or that all arguments not mentioned elsewhere in the opinion are being so rejected. We know that issues not raised on direct appeal and not deemed “fundamental” are considered waived, but we do not know, from the
Woodard
opinion, whether the
Grigsby
claim is “fundamental” for this purpose.
We therefore turn, just as the District Court did, to other opinions of the Supreme Court of Arkansas under Rule 37. The opinion on which the District Court largely relies, 567 F.Supp. at 1570, is
Hulsey v. State, supra,
268 Ark. 312, 595 S.W.2d 934, decided about a year before
Woodard.
There, the Supreme Court warned that Rule 37 is not designed “to serve as a substitute for appeal.”
Id.
at 313, 595 S.W.2d at 935. It went on, though, to consider a
Witherspoon
claim that had not been presented on direct appeal. The opinion almost says in so many words that a
Witherspoon
claim is the equivalent of a claim that a judgment is “void,” the sort of claim expressly covered by Rule 37. It then rejected Hulsey’s
Witherspoon
claim on its merits.
Id.
at 313-14, 595 S.W.2d at 935-36. The District Court has read
Hulsey
as a holding that
Witherspoon
claims are “fundamental” for Rule 37 purposes,
and we agree.
Hulsey’s original Rule 37 petition did not make a
Grigsby
argument. He sought to raise that point in a petition for rehearing filed after the Supreme Court’s original opinion in
Hulsey v. State, supra.
This petition was denied with an opinion,
Hulsey v. State,
268 Ark. 312, 315, 599 S.W.2d 729,
cert. denied,
449 U.S. 938, 101 S.Ct. 337, 66 L.Ed.2d 161 (1980). The District Court reads this opinion as rejecting Hulsey’s
Grigsby
argument on procedural grounds,
namely, that it is not a “fundamental” attack on the judgment of conviction. At this point our own. reading of the Supreme Court of Arkansas’s statements diverges from that of the District Court. The Court’s opinion on rehearing in
Hulsey
does not mention
Grigsby
at all. There is a general discussion of various constitutional arguments, considered as a group, and the Court does say that “it is simply too late for the present contentions to be raised.”
Id.
at 315-B, 599 S.W.2d at 730. We do not read this statement as a holding that
Grigsby
objections are not “fundamental,” for two reasons. First, the
Grigsby
contention was not one of the contentions the court was referring to in this passage, since the passage expressly identifies these contentions as those raised in the original Rule 37 petition, and the
Grigsby
contention was not so raised. Second, the same paragraph in which the arguments are described as being raised too late also labels them “frail and insubstantial.”
Id.
at 315-B, 599 S.W.2d at 730. The procedural ground, thus, is not given as an independent basis of decision, a basis that would have been adequate to support the Court’s judgment even if the constitutional claims had been, on their merits, both strong and substantial.
Furthermore, to read
Woodard
and
Hul-sey
as holding that
Witherspoon
is “fundamental,” but
Grigsby
is not, would create a distinction for which, in our view, no good reason can be given. In Part I of this opinion, we explained the reasons for our holding that
Witherspoon
and
Grigsby
are indistinguishable for retroactivity purposes. Much the same reasoning applies here. Both cases concern the impartiality of the trier of fact. Few, if any, rights are more fundamental than the right to an impartial jury. Indeed, if anything,
Grigs-by
may be more “fundamental” than
With-erspoon:
the latter argument goes only to the sentence, while a
Grigsby
point, if successful, invalidates the conviction itself.
In
Hulsey v. Sargent, supra,
550 F.Supp. at 184, the habeas case brought by Hulsey after his state remedies were exhausted, the District Court gave its reasons for thinking that the Supreme Court of Arkansas might validly distinguish between
Witherspoon
and
Grigsby
in the present context:
Although both
Grigsby
and
Witherspoon
raise the question of an impartial jury, they are distinctly different. A finding that one issue is “fundamental” for these purposes does not necessitate a finding that the other is as well. This is especially true since the Supreme Court in
Witherspoon
established a
per se
rule requiring the invalidity of the death sentence in certain circumstances, whereas it only invited the petitioner to make a record in an attempt to prove the
Grigs-by
claims.
We respectfully disagree with this reasoning. It views the two issues as though they were frozen in time on the day
With-erspoon
was handed down. At that time, of course, the
Witherspoon
rule was fixed, while
Grigsby
was not. But the proper perspective from which to determine whether a question is “fundamental,” it seems to us, must include the assumption that the question will be answered in favor of the party asserting it. A
Witherspoon
claim, if successful, results in the setting aside of a sentence, because the jury that imposed it was not impartial. Just so, a
Grigsby
claim, if successful, results in the setting aside of a verdict of guilty, because the jury that returned it was not impartial. The fact that the
Grigsby
doctrine was inchoate when
Witherspoon
was decided is beside the point.
The real question, in any case, is not whether we think
Grigsby
is just as fundamental as
Witherspoon,
or whether the District Court thinks so, but what the Supreme Court of Arkansas would say on the point. The label “fundamental” is a state-law category: it refers to those issues that can be raised in a Rule 37 petition, notwithstanding the fact that they could have been but were not raised at trial or on direct appeal. The Supreme Court’s opinions in
Woodard
and
Hulsey,
as we read them, do not answer this question directly. But in
Neal v. State,
270 Ark. 442, 605 S.W.2d 421 (1980), the Supreme Court did face the question head on. The Court said:
Neal and his attorneys also seem to have a total misconception of the purpose and role of Rule 37. It is not meant to function as a substitute for appeal, as a method of review of mere error in the conduct of the trial, or as a second opportunity to petition for a rehearing.
Hul-sey v. State,
268 Ark. 312, 595 S.W.2d 934 [(1980)];
Austin v. State,
264 Ark. 318, 571 S.W.2d 584 [ (1978) ];
Clark v. State,
255 Ark. 13, 498 S.W.2d 657 [ (1973) ]. It is not intended to permit the petitioner to again present questions which were passed upon on direct appeal.
Hulsey v. State,
supra. Nor does it permit a petitioner to raise questions which might have been raised at the trial or on the record on direct appeal, unless they are so fundamental as to render the judgment void and open to collateral attack.
Hulsey v. State,
supra.
One point that was raised and disposed of on the first appeal was that jurors were excused by the trial court in violation of
Witherspoon v. Illinois,
391 U.S. 510, 88 S.Ct. 1770, 20 L.Ed.2d 776 (1968). We considered this question on direct appeal and held against appellant in
Neal v. State,
259 Ark. 27, 531 S.W.2d 17 [ (1975) ], so this point is not a ground for relief under Rule 37. The basic question raised [now] is not different, but petitioner seeks to make the argument that a “death qualified” jury is per se unconstitutional, relying upon
Grigsby v. Mabry,
483 F.Supp. 1372 (E.D.Ark.1980), now pending on appeal in the Eighth Circuit Court of Appeals. With all due respect, we simply do not agree with the theory underlying the federal district court’s opinion. If jurors who would not impose the death penalty under any circumstances are not excluded for cause, imposition of the death penalty would be haphazard. We adhere to the views we have expressed in
Grigsby v. State,
260 Ark. 499, 542 S.W.2d 275 [ (1976) ] and
Vena-ble v. State,
260 Ark. 201, 538 S.W.2d 286 [(1976)]. These views are fortified by the decision of the United States Supreme Court in
Lockett v. Ohio,
438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978), where arguments similar to those advanced by petitioner here were rejected.
270 Ark. at 447-48, 605 S.W.2d at 424.
As this passage makes clear, Neal had unsuccessfully made a
Witherspoon
argument on direct appeal. Then, on collateral attack, he raised the
Grigsby
argument. The Supreme Court stated that “[t]he basic question raised is not different,”
id.
at 447, 605 S.W.2d at 424, and plainly rejected the
Grigsby
argument on its merits. The District Court characterized this discussion as “cursory” and “gratuitous,”
Hulsey v. Sargent,
550 F.Supp. at 184, and it may be, in the sense that it was brief and would not have been necessary at all if the Supreme Court had chosen to classify
Grigsby
as non-fundamental and reject the argument on procedural grounds. But that is not at all what the Court did, in our opinion. It rejected the
Grigsby
argument on the merits, even though it could have been but was not raised on direct appeal.
In sum, the Supreme Court’s opinion in
Woodard
does not say whether
Grigsby
is “fundamental” for Rule 37 purposes, but
Neal
says that it is.
Neal
was decided about nine months before
Woodard.
The
Woodard
opinion perhaps omits any discussion of
Grigsby
because the Court had so recently and so clearly stated how much it disagreed with the
Grigsby
theory. The Supreme Court denied Woodard’s petition, thus necessarily rejecting his
Grigsby
argument. From
Neal
we infer that this rejection was a holding on the merits. Therefore, the point is open on habeas under
Ulster County
(in which, incidentally, the state court’s discussion of the merits of the federal question seems to have been no less cursory than that in Neal).
III.
We have held that
Grigsby
applies to this case and that no procedural default in the state courts bars Woodard from raising it. Under
Grigsby
the presence or absence of
prejudice in the particular case is beside the point. At 242.
Because Woodard was not tried by an impartial jury within the meaning of the Sixth and Fourteenth Amendments as interpreted by this Court in
Grigsby,
he is not being lawfully imprisoned. The judgment of the District Court is reversed, and the cause remanded to that court with directions to grant the writ of habeas corpus, unless the State, within such reasonable time as the District Court may fix, commences proceedings to retry Woodard before a validly selected jury.
It is so ordered.