OPINION
BIUNNO, District Judge.Moore attacks his conviction on a plea of guilty on a drug charge on two major *1246grounds. One is that when his plea was accepted, he was not aware of the statutory requirement for a term of mandatory special parole to follow any custodial sentence imposed. 21 U.S.C. § 841(b). The other is that this provision for mandatory special parole is unconstitutional. The proceeding is one under 28 U.S.C. § 2255.
After initial review, the court noted several related questions and asked the parties to express their views on them. One was whether Moore’s counsel had informed him of the requirement, aside from the colloquy at plea acceptance.
Another was whether, if Moore proved to be entitled to withdraw his plea of guilty to one count, the other count dismissed at sentence in accordance with plea negotiations should be reinstated.
The third was whether unenforceability of the provision for special parole had the consequence of establishing that his plea was properly accepted but rendered the sentence illegal.
Following the entry of a memorandum, and then an opinion and order in the matter, an evidentiary hearing was conducted.
The requirement that a defendant charged with an offense in violation of the Comprehensive Drug Abuse Prevention and Control Act of 1970 (21 U.S.C. § 801 et seq.), who may be subject to one or another of the provisions for mandatory special parole, must be aware of that potential consequence was first spelled out in Roberts v. U. S., 491 F.2d 1236 (C.A. 3, 1974), decided January 28, 1974.
Moore had pleaded guilty on October 13, 1972 (15 months before Roberts), and was sentenced on July 27, 1973 (7 months before Roberts), Cr. No. 845-71.
The record at plea shows that the trial judge addressed Moore personally, and made the required findings. Any awareness of the special parole aspects is by inference or implication, rather than explicitly, as noted in the prior rulings.
Moore was represented by counsel and was asked if he had discussed the plea with him; he had. He was asked if counsel had answered all his questions; he had. Since the mandatory special parole provision is set out in the same section Moore was charged with violating and to which he pleaded guilty, counsel could hardly have overlooked it. The inference of awareness is clear and is a proper one.
However, when the question was raised in this proceeding, Moore responded that he was not informed of the provision by counsel. If Moore were taken at his word, the obvious inference could not be drawn.
In those circumstances, if Roberts applies Moore would have the right to withdraw his plea, aside from the other questions discussed later.
The United States meets this reasoning with the claim that Roberts should not be applied retroactively to defendants who had pleaded and were sentenced before the decision in Roberts came down. Heavy reliance is placed on Brown v. U. S., 508 F.2d 618 (C.A. 3 1974), decided December 31, 1974 and amended January 24, 1975.
Brown does contain detailed discussion of the principles and considerations applicable to the question of retroactivity of court decisions. But there are three separate opinions, one by each member of the panel (a majority, a concurring and a dissenting opinion), totalling more than 27 pages, literally bristling with references, excerpts and analyses of many other decisions.
The opinions represent herculean efforts to distill a rational pattern of guides to resolve retroactivity questions. Yet, after multiple reviews of the three opinions as well as of the major decisions of the Supreme Court of the United States on the question, the reader senses that the cases display the characteristics of a Brownian movement more than they do a discernable pattern or matrix.
*1247In principle, retroactivity in the law is a repugnant characteristic to begin with. In the original Constitution, before annexation of the Bill of Rights, the enactment of ex post facto laws was forbidden. U.S.Const., Art. I, § 9, cl. 3. The earliest decision on this was Galder v. Bull, 3 U.S. 305, 3 Dall 386, 1 L.Ed. 648 (1798), and there has been little, if any, deviation from the principles it laid down in regard to criminal and penal statutes.
Aside from criminal and penal statutes, retroactivity is often not barred by paramount law; but the absence of constitutional restriction in no way moderates the repugnant and offensive nature of retroactivity. It smacks of fiat, of arrogance and of self-awarded infallibility.
Retroactivity is difficult and controversial enough when it comes about from the declarations of legislatures or of executive (administrative) agencies. It is at its most controversial when it comes about from judicial decisions.
This is because the judicial branch is widely understood to be engaged in applying the law as it stands; yet new decisions that “make” law pretend to be no more than “discoveries” of what the law has been all along.
Since there are no external and objective means for testing and proving such assertions, they are essentially subjective declarations which draw on nothing more than the authority of the tribunal issuing them for their force and effect.
Should Roberts be applied retroactively? In principle, it should not. In principle it should apply to Mr. Roberts, who succeeded in persuading the court of the merits of his claim, and it should apply to all defendants whose pleas of guilty to the same or similar statutes were tendered and accepted on or after a date soon after Roberts came down, but long enough later for it to have been disseminated and adjustments made to existing procedures. This implies a period of some two to four weeks after the mailing of the slip opinion.
This is the basic principle: that Roberts is necessarily binding on the parties to that case, and it should be binding on like cases occurring thereafter. It should not provide occasion to reopen earlier like eases already disposed of and closed. To do otherwise, i. e., to apply Roberts retroactively, is to raise a far broader question of considerable significance not discussed or dealt with in Roberts or in other cases of similar nature.
Roberts is not alone. The passing years have brought forth many new rulings that have altered the process of tendering and accepting a plea of guilty. The string is now of such length as to raise the question whether any judicial economy can be achieved by acceptance of a plea.
Ideally, sentences should be the same whether conviction follows a plea or a verdict; no premium ought to be placed on the option to stand trial. But since the federal courts do not engage in sentence negotiations (as distinguished from plea negotiations), any defendant sentenced to more than a token penalty will be motivated to try to withdraw his plea on any theory he can find. Most are based on retroactive application of new decisions long after sentence.
If a court decision is intended to be retroactive, it should say so. When the legislative branch enacts a law, it is properly regarded as prospective only, from its effective date forward. Retro-activity of statutes usually rests on an explicit provision. Why should courts not have the same consideration for those who may be affected by its ruling? Silence on retroactivity, with the risk that it may exist, does a great disservice. The burden of uncertainty on the trial courts is not a consideration, great as it may be.
The wholesale, retroactive reopening of closed criminal cases by allowing withdrawal of a guilty plea so that the defendant may stand trial has vast implications for the judicial system. What was a closed case becomes an open case which, because of its age and the incar*1248ceration of the defendant, must be given high priority. This inevitably disrupts the management of the trial calendar.
How old cases are to be reopened and tried, by addition to the aggregate new work flow, in light of the demands of the Speedy Trial Law,1and the new criminal and civil jurisdiction (without regard to diversity or amount) enacted by the new Real Estate Settlement Procedures Act,2 is difficult to assess beyond saying that the effect will not be good.
Even without these new laws there is already an indefensible volume of diversity cases not worth $10,000. in controversy, and allegedly “civil rights” cases that involve routine tort or breach of contract claims having no federal aspect, largely stimulated by the fact that the filing fee in federal court is only $15., while it is $60 in the N.J. courts.
There are also too many cases, tortured and shaped into apparent challenges to the constitutionality of state laws and seeking injunctions, which are so tortured and shaped for the sole purpose of gaining a direct appeal of right to the Supreme Court of the United States. See 28 U.S.C. § 2281, § 2284; 28 U.S.C. § 1253.
The existing and potential deluge threatens now to destroy the functional effectiveness of the judicial branch, both trial and appellate. But this is not the main concern.
The burden on some unknown number of defendants, their victims, the witnesses, the attorneys for both sides, and the law enforcement agencies, is vast by comparison. A simple rule is well established on the civil side, that costs follow the judgment unless otherwise stated. F.R.Civ.P. 54(d). As an equal branch of government and source of law, the judiciary should simply make it clear that decisions are prospective only unless expressly stated to be retroactive. Some courts have done this. See State v. Krol, 68 N.J. 236, at 267, 344 A.2d 289 (1975); State v. Johnson, 68 N.J. 349, at 354, 346 A.2d 66, at 68 (1975).
If courts are to “make” law, they should adhere to the same widely understood and accepted convention as a legislature does. Whether a decision is retroactive or not is not as important a question as knowing whether it is or not.
All are presumed to know the law, but how can anyone have a fair chance if the question of retroactivity is left behind obscuring clouds? Any simple, easily applied rule, no matter how arbitrary, is preferable to the uncertainty of philosophical debate that goes unresolved. The modern world does not have the time to argue the number of angels that can dance on the point of a pin.
If this court were free to do so, it would rule that Roberts is not retroactive. The language of Roberts, however, is that it amounts to no more than an obvious application of Berry v. U. S., 412 F.2d 189 (C.A. 3, 1969), and that it did no more than equate ineligibility for parole (as was the case in Berry) with mandatory special parole to follow any jail sentence (as was the case in Roberts). As this court reads Roberts, it is bound to apply it to Moore even though doing so offends the principle. If this reading be wrong, that determination must be made at a higher level.
So much having been said, the next question is whether Moore should be allowed to withdraw his guilty plea to the one count and stand trial, or whether the withdrawal of the plea should be accompanied by a reinstatement of the other count that was dismissed at sentence pursuant to plea negotiations.
On this question there can be no serious doubt. There was no judgment until Moore was sentenced. The United States honored its commitment to re*1249quest dismissal of the other count at sentence, and it was dismissed.
The dismissal was conditioned on the sentence and the judgment. Withdrawal of the plea, if allowed, amounts to setting aside the judgment. Common sense requires that if the judgment be set aside, other action that was conditioned on the judgment should be set aside as well. To do otherwise would be to magnify the repugnancy of a retroactive application of Roberts.
Moore objects that the two counts are related to the same transaction. That may be so, but any ruling in that regard would need to await trial. The evaluation of evidence by a jury cannot be forecast. A jury might find Moore guilty of the count dismissed-, and not guilty of the one to which he pleaded. Moore’s negotiations may have been aimed at tendering a plea to the count he felt might carry a reasonable doubt, in exchange for dismissal of a count against which he could not defend. If the United States must be held to the agreed terms of plea negotiations, Moore must be held to them, too. Any withdrawal of the guilty plea must be accompanied by reinstatement of the dismissed plea. Justice must be even-handed and impartial.
Beyond- that, dismissals in cases like Moore’s are pursuant to F.R.Crim.P. 48(a), and it is well-established that such dismissals are without prejudice unless expressly stated to be with prejudice. See, for example, 17. S. v. Ortega-Alvarez, 506 F.2d 455 (C.A. 2, 1974); U. S. v. Dobbs, 506 F.2d 445 (C.A. 5, 1975); U. S. v. Fox, 130 F.2d 56, at 58 (C.A. 3, 1942).
Since the discovery of the problem in this case, this court is taking the precaution, in cases where a defendant pleads guilty and is sentenced on one or more counts, under plea arrangements calling for dismissal of other counts or other indictments, to qualify the leave granted by explicitly stating that the dismissal is on limitation, i. e., only so long as the plea and sentence imposed remain in full force and effect.
The constitutional issue is one on which no court has squarely ruled in the context of this ease. The handful of reported cases have considered the issue in circumstances that seem not to have required more than a casual treatment. E. g. U. S. v. Simpson, 481 F.2d 582 (C.A. 5, 1973) cert. den. 414 U.S. 1095, 94 S.Ct. 728, 38 L.Ed.2d 553; U. S. v. Martinez; 481 F.2d 214, (C.A. 5, 1974) cert. den. 415 U.S. 931, 94 S.Ct. 1444, 39 L.Ed.2d 489.
The context here is different, for in this circuit not only does Roberts require that a defendant be aware of the “outer limits” of the penalty, but Kelsey v. U. S., 484 F.2d 1198 (C.A. 3, 1973) does not permit a plea -to be accepted or to stand in a case where the outer limits described go beyond the actual limits permitted.
In the case of the mandatory special parole clause there is no rational way in which both Roberts and Kelsey can be complied with. This is because the statute sets only a minimum (actually, a set of minimums) and no “outer limit” of any kind, and no norm or standard that can be used to make a defendant understandably aware of the “outer limits.”
Should a defendant charged under this law be told that if given any custodial sentence, there will be imposed a mandatory special parole of at least 2, 3, 4, 6 or 9 years (depending on the classification of the dangerous substance, of any prior conviction, or of the distribution to a minor) and for as- long as 99 years, or for. life? The statute is silent. If what he is told is too long, the plea cannot stand under Kelsey. If what he is told is too short, the plea cannot stand under Roberts. The problem is like that of Procrustes’ Bed. See, “Lincoln Library of Essential Information”, p. 307 37th Ed., Frontier Press, Columbus, Ohio.
*1250It must be noted that the requirement is for “special parole”, not for probation. If the statute called for probation to follow imprisonment, 18 U.S.C. § 3651 would limit the period of probation to five years. This cannot have been the intent of Congress, since a second offender must be put on at least double the applicable minimum of 2 or 3 years of special parole, and a distributor to a minor must be sentenced to triple those mínimums. And if he violates parole, the prisoner must serve the special parole term, Roberts, supra.
Even when a youth offender is sentenced to an indefinite term under 18 U.S.C. § 5010(b), there are standards for both conditional and unconditional release specified in 18 U.S.C. § 5017(c).
These being the circumstances, there is no possible way in which a trial court can establish that a defendant subject to a special parole term understands the consequences of his plea. Nor does a court, at- plea, have reliably available all the information that will control the minimum. This, in turn, would make it impossible to accept a plea in such cases.
The result is that the provision for mandatory special parole is unenforceable as in conflict with paramount law. Every defendant has the right to tender a plea and to have that plea accepted if the requisite findings can be made. Since the findings cannot be made, the provision is necessarily unenforceable.
The next question is whether the provision is severable. There cannot be any doubt that it is. The present statute was enacted, in part, to meet widespread criticism of the mandatory minimum jail sentences of the prior law which it replaced, and Congress obviously included the special parole clause in substitution for the minimum jail sentence. It hardly intended that the serious offenses covered by the new law should go wholly unpunished.
The Congress can, of course, easily correct the defect if it wishes, by stating some particular maximum limit on any single special parole term (such as double the minimum), and on the aggregate of all such terms (such as the remaining life expectancy of the defendant after expiration of all custodial sentences).
Finally, the court finds as a fact that Moore was both advised and aware of the mandatory special parole provision. The evidence at the hearing, as well as the records and files show this beyond question.
While Moore’s attorney, who appeared at plea and at sentence could not explicitly recall when he told Moore about this provision, he was certain he had explained it because it is spelled out in the very section of the act that Moore was charged with. He told him it applied only if there were a jail sentence. His certainty is confirmed by his testimony that on sentence day, while waiting to be reached, two other sentences involving the mandatory special parole provision were imposed, and the attorney does explicitly recall nudging Moore and reminding him that this was the provision they had talked about. The minutes of the court on that date (July 27, 1973, Exh. G-1) confirm that the two others were so sentenced (Michael Hal-lock and Thomas Bertsch, both on Cr. No. 429-72).
Moore’s testimony to the contrary is not worthy of belief, and the court does not believe him. His explanation that he thought the 3 years of special parole meant that he would be paroled in 3 years is not credible. First, he would be eligible for parole after 3 years and 4 months (on a 10 year sentence) in any event, 18 U.S.C. § 4202. Second, as the sentence transcript shows, Moore was sentenced for “a term of ten years to be followed by three years special parole” (emphasis added). This is the expression invariably used by the court, and Moore heard it twice before he heard it on his own sentence.
Beyond that, Moore twice applied for reduction of sentence under F.R.Cr.P. 35' (before the present motion) and never *1251made any mention of or complaint about the 3 years of special parole. The court finds that Moore was in fact fully aware of and fully understood the special parole aspect, (to the extent it can be) but was quite unhappy about the 10 years, which he cannot challenge.
Accordingly, from what has been decided, Moore cannot withdraw his plea. But, since the provision has been found unenforceable, it follows that the sentence imposed was necessarily illegal to the extent that it included the 3 year special parole. Under F.R.Crim.P. 35, that sentence can be corrected at any time, and under 28 U.S.C. § 2255, the presence of Moore for the correction of sentence is not required.
A corrected judgment conforming to this opinion is entered at the same time as this opinion.
. P.L. 93-619 (1975) ; 18 U.S.C.A. § 3161-3174.
. P.L. 93-533 (1974) ; 12 U.S.C.A. § 2601-2616.