This appeal is before the court upon the remand of the Supreme Court, Bowen v. United States, 413 U.S. 915, 93 S.Ct. 3069, 37 L.Ed.2d 1038 (1973), vacating, 462 F.2d 347 (9th Cir. 1972).

Bowen was convicted of smuggling and transporting marijuana and of possessing depressant and stimulant drugs. The evidence of the violations was discovered during a routine search for illegal aliens of a camper truck at a permanent border-patrol checkpoint on California State Highway 86 approximately 36 air miles and 49 highway miles north of the Mexican border. Highway 86 is a principal route from Mexicali to Los Angeles by way of Indio and Riverside.

On June 21, 1973, the Supreme Court held in Almeida-Sanchez v. United States, 413 U.S. 266, 93 S.Ct. 2535, 37 L.Ed.2d 596 (1973), rev'g, 452 F.2d 459 (9th Cir. 1971), that border-patrol agents on roving patrol cannot stop and search cars pursuant to 8 U.S.C. 1357(a) and 8 C.F.R. 287.1 without probable cause or warrant.

Two separate issues are presented here: (1) How does Almeida-Sanchez affect searches conducted at a fixed checkpoint? (2) If fixed-checkpoint searches, as well as roving-patrol searches, are included within the ambit of the Almeida-Sanchez ruling, should that ruling be applied to fixed-checkpoint searches conducted by border-patrol agents prior to June 21, 1973, in cases pending on appeal on that date?

For the reasons set forth in Part I of the opinion of the majority of the court here, we hold that the rule announced by the Supreme Court in Almeida-Sanchez does apply to searches at fixed checkpoints. However, for the reasons set forth in Part II, we also hold that Almeida-Sanchez will not be applied to fixed-checkpoint searches conducted prior to June 21, 1973.

The judgment of conviction is affirmed.

According to the Supreme Court in Almeida-Sanchez v. United States, 413 U.S. 266, 93 S.Ct. 2535, 37 L.Ed.2d 596 (1973), the government has been conducting three types of alien searches pursuant to 8 U.S.C. 1357(a) and 8 C.F.R. 287.1: (1) searches at 'permanent checkpoints * * * maintained at certain nodal intersections;' (2) searches at 'temporary checkpoints * * * established from time to time at various places'; and (3) searches carried out by 'roving patrols.' 413 U.S. at 268. The government argued in Almeida-Sanchez that all these searches conducted 'within a reasonable distance from any external boundary,' 8 U.S.C. 1357(a)(3), could be considered border searches, and thus be carried out with neither a warrant nor probable cause. See Carroll v. United States, 267 U.S. 132, 154, 45 S.Ct. 280, 69 L.Ed. 543 (1925).

The search condemned in Almeida-Sanchez was of the 'roving patrol' type. It was conducted 25 miles north of the Mexican border, on a California eastwest highway that lies at all points at least 20 miles north of the border. 413 U.S. at 267-268, 273. The search of Bowen's camper, however, was a fixed-checkpoint search, a type of search reserved from the Almeida-Sanchez decision. The checkpoint, on California State Highway 86, was between the major population centers of the Imperial Valley and Indio.

The opinion1 in Almeida-Sanchez, delivered by Mr. Justice Stewart, leaves little doubt that traditional Fourth Amendment standards apply to fixed-checkpoint searches as well as to roving-patrol searches. Early in the opinion, after listing the three types of surveillance conducted by the Border Patrol along inland roadways and noting that the government argues that 'in all of these operations * * * the agents are acting within the Constitution when they stop and search automobiles without a warrant, without probable cause to believe the cars contain aliens, and even without probable cause to believe the cars have made a border crossing,' the Court stated that 'the only asserted justification for this extravagant license to search is 287 of the Immigration and Nationality Act, 66 Stat. 233, 8 U.S.C. 1357(a) * * *.' 413 U.S. at 268.

Moreover, the government in Almeida-Sanchez sought to justify roving-patrol searches on the basis of 8 U.S.C. 1357(a)(3) and 8 C.F.R. 287.1(a)(2). Here, the government seeks to justify the fixed-checkpoint search by reference to the same statute and regulation. But, when the Supreme Court held that this statute and regulation could not exempt searches carried out pursuant to them from traditional Fourth Amendment scrutiny, see 413 U.S. at 272, the government's statutory justification for fixed-checkpoint searches as well as for roving-patrol searches vanished.

Finally, at the very end of its opinion, 413 U.S. at 274-275, the Court quoted from Carroll v. United States as follows:

'* * * It would be intolerable and unreasonable if a prohibition agent were authorized to stop every automobile on the chance of finding liquor and thus subject all persons lawfully using the highways to the inconvenience and indignity of such a search. Travelers may be stopped in crossing an international boundary because of national self-protection reasonably requiring one entering the country to identify himself as entitled to come in, and his belongings as effects which may be lawfully brought in. But those lawfully within the country, entitled to use the public highways, have a right to free passage without interruption or search unless there is known to a competent official authorized to search, probable cause for believing that their vehicles are carrying contraband or illegal merchandise * * *.' 267 U.S. at 153-154.

It would be strange indeed for the Court to quote this language if it meant to leave intact the government's asserted right to establish fixed checkpoints anywhere within 100 miles of the border. Surely, searches at these checkpoints, like searches by roving patrols have the effect of violating the 'right to free passage without interruption or search' of 'those lawfully within the country * * *.' Such searches, therefore, must meet constitutional standards regardless of their utility in carrying forward the difficult mission of the Border Patrol.

To be sure, Mr. Justice Powell in his concurrence and Mr. Justice White in his dissent both correctly pointed out that Almeida-Sanchez did not present a question of a fixed-checkpoint search. See 413 U.S. at 275-276, 288. Nonetheless, these disclaimers do not override clear indications in the opinion of the Court that any distinction between fixed and movable checkpoints will be meaningless unless the distinction can be based upon reasoned Fourth Amendment considerations.

The government argues that there are, in fact, significant constitutional differences between roving patrols and fixed checkpoints. First, since fixed checkpoints often involve a stop and inspection of every car passing through them, they provide much less opportunity for the unfettered discretion of the police officer that was condemned in Almeida-Sanchez. See 413 U.S. at 268. Second, being stopped on a lonely road at night in a sparsely populated part of the country (Almeida-Sanchez) is more burdensome to the traveler than a stop at an identified and lighted checkpoint (Bowen). The government contends once again that the Constitution does not forbid all searches, but only 'unreasonable' ones. In support of the validity of arguably 'reasonable' fixed-checkpoint searches, the government cites a number of judicial decisions upholding roadblocks established for the purpose of checking drivers' licenses and registrations. See, e.g., United States v. Croft, 429 F.2d 884, 886 (10th Cir. 1970); People v. Washburn, 265 Cal.App.2d 665, 71 Cal.Rptr. 577, 581 (2d Dist. 1968); State v. Smolen, 4 Conn.Cir. 385, 232 A.2d 339 (App.Div.), pet. for certification for appeal denied, 231 A.2d 283 (Conn.1967), cert. denied, 389 U.S. 1044, 88 S.Ct. 787, 19 L.Ed.2d 835 (1968); State v. Severance, 108 N.H. 404, 237 A.2d 683 (1968).2

We agree with the government that a fixed-checkpoint search that does not entail significant delay is probably less offensive than a roving-patrol search. Being asked to stop at a fixed checkpoint is not frightening to a seasoned traveler. Being flagged over to the side of the road by a roving patrol might be traumatic. Also, an officer on roving patrol probably has more discretion in deciding which cars to stop than one stationed at a fixed checkpoint, although the difference might be less than the government contends. Since not all vehicles passing through a checkpoint are stopped, and since not all vehicles stopped are searched, the officer at the checkpoint still retains a good deal of discretion to 'single out' some travelers for stops or intrusive searches.

Nonetheless, even conceding that a fixed-checkpoint search might be less of an imposition on domestic travelers than a roving-patrol search, we are able to find nothing in the opinion of the Court in Almeida-Sanchez which suspends Fourth Ameida-Sanchez which suspends Fourth Amendment standards in dealing checkpoints.

Moreover, the government's reliance on judicial decisions upholding automobile stops for the purpose of checking drivers' licenses is misplaced. Although the government has cited only roadblock stops, there is a paralled line of decisions upholding roving-patrol stops to check for valid license and registration. See, e.g., Lipton v. United States, 348 F.2d 591 (9th Cir. 1965); State v. Ream, 19 Ariz.App. 131, 505 P.2d 569 (Div. 1, Dept. B, 1973); Leonard v. State, 496 S.W.2d 576 (Tex.Cr.App.1973). In other cases, courts did not even find the fact that the stop may have been at a roadblock rather than incident to a roving patrol significant enough to mention. See, e.g., United States v. Lepinski, 460 F.2d 234, 237 (10th Cir. 1972); Myricks v. United States, 370 F.2d 901 (5th Cir.), pet. for cert. dismissed, 386 U.S. 1015, 87 S.Ct. 1366, 18 L.Ed.2d 474 (1967). Hence, since this line of vehicle-registration-check decisions was as relevant in Almeida-Sanchez as here, we do not find these decisions to be an appropriate means for distinguishing Bowen's search from that of Almeida-Sanchez.

What is more, the rationale for the drivers'-license decisions will not support the government's position here. For example, in Lipton v. United States, supra, in which this court upheld a stop by a motorcycle police officer of a youth driving an automobile, we reasoned that there was no way for a police officer to determine that a driver had a valid license permitting him to operate a motor vehicle other than by stopping him and asking him to produce his license. We noted:

'* * * A contrary holding would render unenforceable the State statute requiring that automobile drivers be licensed.' 348 F.2d at 593. We are not persuaded that laws prohibiting illegal immigration will be rendered similarly unenforceable should we deny to the government the power to stop and search automobiles, without probable cause or warrant, at fixed checkpoints.

We hold, then, that fixed-checkpoint searches, like roving-patrol searches, even though conducted within a 'reasonable distance' from the border, are not necessarily exempt from the traditional Fourth Amendment requirement of a warrant or probable cause. This holding, however, merely shifts the focus of our inquiry. The opinion in Almeida-Sanchez does not require that a border search, to be constitutional, be at the border itself; rather, a legitimate border search may also be conducted 'in certain circumstances' at the border's 'functional equivalents.' 413 U.S. at 272. The search conducted in the present case was obviously not at the border itself; nor was it at a 'functional equivalent' of the border.

The 'function' of a border checkpoint is to regulate border crossings. Thus, in attempting to clarify what would constitute a 'functional equivalent' of the border, the Court in Almeida-Sanchez offered two examples:

'* * * For example, searches at an established station near the border, at a point marking the confluence of two or more roads that extend from the border, might be functional equivalents of border searches. For another example, a search of the passengers and cargo of an airplane arriving at a St. Louis airport after a nonstop flight from Mexico City would clearly be the functional equivalent of a border search.' 413 U.S. at 272-273.

These examples are then contrasted with the search conducted in Almeida-Sanchez:

'* * * The search of * * * (an) automobile by a roving patrol, on a California road that lies at all points at least 20 miles north of the Mexican border, was of a wholly different sort * * *.' 413 U.S. at 273.3

In other words, if a search takes place at a location where virtually everyone searched has just come from the other side of the border, the search is a functional equivalent of a border search. In contrast, if a search takes place at a location where a significant number of those stopped are domestic travelers going from one point to another within the United States, the search is not the functional equivalent of a border search. One need only contemplate the volume of domestic travel between Buffalo and Rochester, New York, to see why a checkpoint between those two cities could not be the functional equivalent of a border checkpoint even though the checkpoint could be less than twenty miles from an international border.

In addition to the two examples of a functional equivalent of a border search provided by the Court in Almeida-Sanchez, other examples may be drawn from two series of decisions of this court. Representative of the first line of authority is Alexander v. United States, 362 F.2d 379 (9th Cir.), cert. denied, 385 U.S. 977, 87 S.Ct. 519, 17 L.Ed.2d 439 (1966). In that case customs officials, acting upon a tip from an informer, placed the defendant's automobile under surveillance when it crossed the border, and kept it under almost continuous watch as it made suspicious movements through the streets of a border city. In holding that a subsequent search by customs officials was properly called a border search, the court stated:

'Where * * * a search for contraband by Customs officers is not made at or in the immediate vicinity of the point of international border crossing, the legality of the search must be tested by a determination whether the totality of the surrounding circumstances, including the time and distance elapsed as well as the manner and extent of surveillance, are such as to convince the fact finder with reasonable certainty that any contraband which might be found in or on the vehicle at the time of the search was aboard the vehicle at the time of entry into the jurisdiction of the United States * * *.' 362 F.2d at 382.4

In a second line of our own cases, this court has treated a search north of the border as the equivalent of a border search, where it appeared with reasonable certainty that the vehicle searched contained either goods which have just been smuggled or a person who had just crossed the border illegally. See, e.g., United States v. Weil, 432 F.2d 1320 (9th Cir. 1970), cert. denied, 401 U.S. 947, 91 S.Ct. 933, 28 L.Ed.2d 230 (1971), in which this court held:

'* * * If customs agents are reasonably certain that parcels have been (a) smuggled across the border and (b) placed in a vehicle, whether the vehicle has itself crossed the border or not, they may stop and search the vehicle. Similarly, if agents are reasonably certain that a person has crossed the border illegally, and has then entered a vehicle on this side of the border, we think, that they may stop and search the vehicle and person. They can assume that he may have brought something with him.' 432 F.2d at 1323.5

In both of these pre-Almeida-Sanchez lines of Ninth Circuit authority, as well as in the two examples offered by the Supreme Court in Almeida-Sanchez, although the search was not conducted precisely at the border, it still was directly related to a recent entry from across a border. See United States v. Almeida-Sanchez, 452 F.2d at 463 (dissenting opinion of Browning, J.), rev'd, 413 U.S. 266, 93 S.Ct. 2535, 37 L.Ed.2d 596 (1973).

It is clear that the search conducted in the present case lacks the vital connection between the vehicle stopped and a reasonable certainty, or even a probability, that it or its contents had recently crossed an international border. The checkpoint was approximately 36 air miles and 49 highway miles north of the Mexican border. Several significant population centers and highways, including Interstate 8, a major east-west freeway that connects the Southwest with the West Coast, lie between the checkpoint and the border. Under these circumstances, border-patrol agents had no reason to believe that virtually all or even most of the cars passing through their checkpoint had recently, or ever, crossed the border. Moreover, there was neither the continuing surveillance from the border nor the dependable intelligence from other sources required to fit this case into the Alexander line of decisions; nor was there the reasonable certainty that the vehicle contained either recently smuggled goods or aliens required under the Weil line. See United States v. Petersen, 473 F.2d 874 (9th Cir. 1973); United States v. Mitchell, 472 F.2d 67, 68 n. 1 (9th Cir. 1973). Hence, the record in this case clearly indicates that the search conducted was not the functional equivalent of a border search. Were the record more equivocal, we would not hesitate to remand the case to the district court for a determination of functional equivalency. Where the record is as clear as it is here, however, we see no need for a remand.6

As its last line of defense, the government argues that fixed-checkpoint searches, even if not the functional equivalent of border searches, should be upheld simply because they are urgently needed. The government's difficulty in detecting and repatriating illegal aliens along our southern boundary needs no new documentation here. The short answer to this argument, however, is that necessity alone cannot override the Fourth Amendment's prohibition against unreasonable searches and seizures. A similar argument was made and rejected in Almeida-Sanchez itself. See 413 U.S. at 293 (dissenting opinion of White, J.). Mr. Justice Powell in his concurring opinion in Almeida-Sanchez suggested that warrants based on area-wide conditions could be employed to resolve the contending interests of law enforcement and Fourth Amendment safeguards. 413 U.S. at 275-285; see generally The Supreme Court-- 1972 Term, 87 Harv.L.Rev. 55, 200-04 (1973). Since the government did not seek such a warrant in this case, we need not now express an opinion on a hypothetical search conducted pursuant to a judicial warrant authorizing searches for a limited time at a specific checkpoint. We refer to Mr. Justice Powell's opinion here merely to suggest to the government that procedures less offensive to the Fourth Amendment than judicially unapproved checkpoint searches might be devised and implemented to supplement its program for enforcing immigration laws.

Our conclusion that Almeida-Sanchez is as applicable to fixed-checkpoint searches as to roving-patrol searches is consistent with that reached by the Court of Appeals for the Fifth Circuit in United States v. Speed, 489 F.2d 478 (5th Cir. 1973). There, the court held that a border-patrol search at temporary checkpoint on a north-south highway approximately 65 to 75 miles north of the Mexican border was neither a border search nor a functional equivalent of a border search and, hence, was unconstitutional. The court commented, 'The distinction between a checkpoint and a roving patrol is not important.' 489 F.2d at 480.

Likewise, the Court of Appeals for the Tenth Circuit has held that a warrantless search, without probable cause, of an automobile at the checkpoint at Truth or Consequences, New Mexico, violates the Fourth Amendment unless a search at that checkpoint could be deemed the functional equivalent of a border search. United States v. King, 485 F.2d 353 (10th Cir. 1973); United States v. Maddox, 485 F.2d 361 (10th Cir. 1973).

We hold here that under the rule announced by the Supreme Court in Almeida-Sanchez the search of Bowen's camper truck violated the Fourth Amendment because the search was not the functional equivalent of a border search and was authorized neither by warrant nor by probable cause.

Circuit Judges MERRILL, BROWNING, DUNIWAY, ELY, HUFSTEDLER and TRASK concur in this majority opinion (Part I).

The first step in deciding whether a case is to have retroactive effect is to apply a threshold test to determine whether the decision establishes a new rule. If it does, we proceed to test whether the new rule is retroactive. If it does not, no such testing is necessary as, by definition, without a new rule, there is no change in the law and the question of retroactivity is immaterial. For purposes of this case, we hold that to constitute a new rule, the decision must either (1) overrule clear past precedent or (2) disrupt a practice long accepted and widely relied upon.7

Under the first alternative of the test, a decision constitutes a new constitutional rule if it overrules clear past precedent. The test does not require, however, that the Supreme Court reverse itself in order for there to be an overruling of clear past precedent. Such a requirement would not only be illogical but would also be violative of the reason for the rule. Only a small number of the appealed federal cases are ever reviewed by the Supreme Court. As a matter of necessity, the Court must leave the day-to-day application of the law to the lower courts. Therefore, law enforcement procedures must be based to a great extent upon circuit court decisions. Because of this necessity, a law enforcement practice may develop and be sanctioned by court approval for many years before it is reviewed by the Supreme Court. Were such a rule is ultimately reversed by the Court, the pronouncement is 'new' simply by virtue of the fact that the people who apply the law on a day-to-day basis have not previously understood the new statement to be the proper rule. Justice Blackmun recognized this principle, noting specifically that the new pronouncement need not overrule a prior Supreme Court case, when he wrote in Gosa v. Mayden, 413 U.S. 665, 673, 93 S.Ct. 2926, 2932, 37 L.Ed.2d 873 (1973):

Although the Court in O'Callahan did not expressly overrule any prior decision, it did announce a new constitutional principle, and it effected a decisional change in attitude that had prevailed for many decades.

Prior to Almeida-Sanchez, there was no indication from any of the courts of appeals that searches at fixed checkpoints were unconstitutional. Rather, the indication was that such searches were constitutional, for 35 of 36 judges who had considered the question in the three circuits involved in enforcing the immigration laws along our Mexican border had upheld immigration stops and searches. Almeida-Sanchez v. United States, 413 U.S. at 298-299 n. 10 (White, J., dissenting).

In addition to the numerous court pronouncements, the statute pursuant to which the Border Patrol acted also constitutes clear past precedent. Since 1952, searches for aliens within a reasonable distance from the border have been authorized by Congress. 8 U.S.C. 1357(a)(3). Obviously, such a statute establishes a legal rule until repealed or declared unconstitutional. Although the Supreme Court did hold that, under the facts of Almeida-Sanchez, that statute cannot make a search conducted by a roving patrol reasonable within the Fourth Amendment, it was the prior law just as much as Wolf v. Colorado, 338 U.S. 25, 69 S.Ct. 1359, 93 L.Ed. 1782 (1949), was the prior law until Mapp v. Ohio, 367 U.S. 643, 81 S.Ct. 1684, 6 L.Ed.2d 1081 (1961). The prior law, be it statutory or case law, guides law enforcement personnel and courts until abrogated. Here, the prior law was, in part, statutory and remained unassailed by the Supreme Court or circuit courts for twenty years. It was then limited by Almeida-Sanchez. Although it is true that statutes have to be measured by the Constitution, a legally enacted statute becomes the law until it is vitiated by a court decision. Where the constitutionality of the statute has been repeatedly upheld by the lower courts, it becomes a clear precedent for law enforcement action. Prior statutory law should be treated no differently from prior case law. Thus, Almeida-Sanchez, as applied to searches at fixed checkpoints, overrules clear past precedent, both statutory and case law.

Under the second alternative of the threshold test, a decision constitutes a new constitutional rule if it disrupts a practice long accepted and widely relied upon. Here too, Almeida-Sanchez measures up as a new pronouncement. For over a decade we have consistently and repeatedly upheld convictions based upon evidence seized during searches made at fixed checkpoints; in none of these cases has the constitutionality of such searches been questioned.8 A law enforcement practice authorized by our court for such an extensive period easily satisfies the second alternative of the test. Under either alternative, Almeida-Sanchez is a new pronouncement.

We, therefore, must determine whether this new rule should be applied retroactively to fixed checkpoints. As the Constitution does not mandate retroactive application, Linkletter v. Walker, 381 U.S. 618, 629, 85 S.Ct. 1731, 14 L.Ed.2d 601 (1965), we must use judicially conceived guidelines to resolve the issue. The tripartite test developed by the Supreme Court in numerous decisions over the last decade was enunciated in Stovall v. Denno, 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967), and recently reiterated in Michigan v. Payne, 412 U.S. 47, 51, 93 S.Ct. 1966, 36 L.Ed.2d 736 (1973). The Stovall test requires that, before we apply new constitutional protections retroactively, we consider the following three criteria:

(a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards.

Stovall v. Denno, 388 U.S. at 297.

Criterion (a) of the Stovall test, the purpose to be served by the new standard, generally dictates that the new pronouncement be applied retroactively only when the pronouncement affects the 'fairness of the triai-- the very integrity of the fact-finding process.' Linkletter v. Walker, 381 U.S. at 639. Since the legality of the search in no way affects the court's ability to determine whether or not the defendant is actually guilty of committing the crime, a pronouncement requiring the suppression of such evidence serves only as a procedural device to deter future improper police conduct. See United States v. Calandra, 414 U.S. 338, 94 S.Ct. 613, 38 L.Ed.2d 561 (1974). Such pronouncements generally should not be applied retroactively. As we held in Williams v. United States, 418 F.2d 159, 162 (9th Cir. 1969), aff'd, 401 U.S. 646, 91 S.Ct. 1148, 28 L.Ed.2d 388 (1971):

The Court in Desist (Desist v. U.S., 394 U.S. 244, 89 S.Ct. 1030, 22 L.Ed.2d 248 (1969)) said the foremost of the three criteria was the first. If the purpose is to deter misconduct of police officers in conducting a search, the new exclusionary rule will not be given retrospective effect because that purpose is not advanced by penalizing conduct that has already occurred. The exclusionary rule in such cases, the Court observed, was a procedural device to curb illegal police action and not a rule affecting the integrity of the process for finding the innocence or guilt of an accused.

Criterion (b) of the Stovall test, the reliance of law enforcement officers on the old standards, is similar to the second alternative of our threshold test. A careful analysis demonstrates that law enforcement personnel justifiedly relied on the prior rule. Since Almeida-Sanchez did not determine the constitutionality of searches at a fixed checkpoint, let alone establish a new rule as to such searches, applying Almeida-Sanchez to fixed checkpoints would overturn our own Ninth Circuit precedents which stretch back over a decade.

In Fernandez v. United States, 321 F.2d 283 (9th Cir. 1963), we were specifically faced with the issue of the constitutionality of the statute and regulations authorizing stops and limited searches at a fixed checkpoint. Appellant Fernandez expressly specified as error:

1. Regulations sanctioning check points 70 miles within the United States are unconstitutional and illegal per se, and are unconstitutional and illegal as applied in this case. 2. The stopping of cars 70 miles within the confines of the United States constitutes unconstitutional and illegal search and is unlawful without probable cause. 3. Evidence discovered during a search following such illegal stopping of a car is 'fruit of a poisoned tree', and must be suppressed.

321 F.2d at 285. We found that the statute and regulations, pursuant to which the checkpoints were operated, were 'clearly constitutional.' After reviewing extensive findings of fact by the district court concerning the necessity of such checkpoints, we concluded that the statute and regulations as applied were 'neither arbitrary nor capricious.' We further held that after the officers had legally stopped the car, their detection of the odor of marijuana constituted probable cause sufficient to justify a search of the car. We have repeatedly reaffirmed Fernandez, not only upholding the validity of the initial stopping of the car, but also upholding reasonable searches for aliens made without probable cause. In Fumagalli v. United States, 429 F.2d 1011 (9th Cir. 1970), after reviewing Fernandez and a number of our other earlier cases, we concluded:

What all of these cases make clear is that probable cause is not required for an immigration search within approved limits but is generally required to sustain the legality of a search for contraband in a person's automobile conducted away from the international borders.

429 F.2d at 1013. Given the number and the explicitness of our prior prouncements,9 it would be hard to find a clearer case justifying reliance on an old standard. Immigration stops and searches at fixed checkpoints, a practice that has extended over a 44 year period, has enjoyed ten years of continued and uninterrupted judicial approval. If the border patrol agents cannot rely upon a statute supported by clear regulations which have repeatedly been upheld by a Court of Appeals with no Supreme Court disapproval, it is difficult to conceive what degree of official pronouncements would be necessary to make their reliance justified. See Adams v. Illinois, 405 U.S. 278, 283-284, 92 S.Ct. 916, 31 L.Ed.2d 202 (1972); Johnson v. New Jersey, 384 U.S. 719, 731, 86 S.Ct. 1772, 16 L.Ed.2d 882 (1966).

What was said in Stovall v. Denno, 388 U.S. 293, 299-300, 87 S.Ct. 1967, 1971, 18 L.Ed.2d 1199 (1967), is equally applicable here:

Today's rulings were not foreshadowed in our cases; no court announced such a requirement . . .. Law enforcement authorities fairly relied on this virtually unanimous weight of authority, now no longer valid, in conducting pretrial confrontation in the absence of counsel.

Criterion (c) of the Stovall test requires that we consider the effect the retroactive application of the new pronouncement would have on the administration of justice. The record before us is inadequate to indicate what the effect would be in requiring probable cause or a warrant for all prior stops and searches made at fixed checkpoints. For example, it is impossible to tell how many illegal aliens, as well as alien smugglers and other criminals, were captured redhanded at fixed checkpoints since their inception in 1929. Nonetheless, we are aware that in fiscal year 1973 alone, there were approximately 55,300 deportable aliens apprehended at traffic checkpoints. United States v. Baca, 368 F.Supp. 398, 407 (S.D.Cal.1973). There can be no doubt that a rule requiring the suppression of all evidence obtained by searches made at fixed checkpoints without a warrant or probable cause, applied retroactively, would create immense problems hindering the administration of justice. It is not inconceivable that thousands of convicted criminals might have to be retried or set free.

Nor do we believe, as has been contended, that Robinson v. Neil, 409 U.S. 505, 93 S.Ct. 876, 35 L.Ed.2d 29 (1973), should lead us to a different conclusion. There, retroactivity was applied to a non-fact-finding aspect (double jeopardy). But the Court clearly noted that it was creating an exception to the Linkletter-Stovall approach similar to that which made the death penalty cases retroactive. The Court explained that prospective application has generally been applied under Linkletter when the rule was directed 'to collateral purposes such as the deterrence of unlawful police conduct . . ..' Id. at 509. They then specifically distinguished rules with collateral purposes from the double jeopardy situation where a defendant could possibly be punished twice for the same crime. The obvious import of Robinson in reference to the case before us is that where the purpose of the rule is the deterrence of unlawful police conduct, the rule will be prospective only.

The only remaining question is the date upon which Almeida-Sanchez would become applicable to searches at fixed checkpoints. Some would argue that there should be at least a limited retroactivity, requiring us to apply the new rule to those cases involving searches at fixed checkpoints that are now on direct appeal. These are the so-called 'pipeline' cases. We reject this approach and hold that Almeida-Sanchez applies only to searches at fixed checkpoints after June 21, 1973, the date of the Almeida-Sanchez decision. The Supreme Court's recent decisions indicate that the pipeline theory does not enjoy majority approval. See Stovall v. Denno, 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967). The Court had precisely that issue before it in Williams v. United States, 401 U.S. 646, 91 S.Ct. 1148, 28 L.Ed.2d 388 (1971), and a majority declined to apply the new rule either to the cases in the pipeline (on direct appeal) or to the cases that were before the Court on collateral attack. Only Justices Brennan and Marshall supported the pipeline theory.

In Michigan v. Payne, 412 U.S. 47, 93 S.Ct. 1966, 36 L.Ed.2d 736 (1973), the Court again adopted limited prospectivity, i.e., only the challenging appellant would benefit from the new rule.10 In Payne, the Court held that the prophylactic limitations established in North Carolina v. Pearce, 395 U.S. 711, 723-726, 89 S.Ct. 2072, 23 L.Ed.2d 656 (1969), would not be applicable to resentencing proceedings that occurred prior to the date of the Pearce decision, even though Payne's appeal was in the pipeline when Pearce was decided. Justice Marshall, dissenting, concluded that 'considerations of fairness rooted in the Constitution (require) that cases in the pipeline when a new constitutional rule is announced must be given the benefit of that rule.' 412 U.S. at 60. None of the other Justices joined in this part of his dissent and Justice Marshall himself admitted that, other than exceptions not applicable in this case, all 'constitutional rules of criminal procedure have been given prospective effect only.' 412 U.S. at 62. He noted that limited retroactivity, as applied in Linkletter (381 U.S. at 622), was an 'anomaly.'11 It would be unwise for us to adopt the pipeline theory when the Court has declined to apply it.

Limited prospectivity is not a new idea in the line of cases dealing with the retroactivity of new rules of criminal procedure. In Johnson v. New Jersey, 384 U.S. 719, 86 S.Ct. 1772, 16 L.Ed.2d 882 (1966), decided only one year after Linkletter, the Court held that Escobedo and Miranda would apply only to cases in which the trial began after the dates of the decisions. One could argue that Johnson does not really deal with the pipeline theory because the case was before the Court on collateral attack. But such an argument would ignore the fact that the Court expressly announced that:

We do not find any persuasive reason to extend Escobedo and Miranda to cases tried before those decisions were announced, even though the cases may still be on direct appeal.

384 U.S. at 733. Limited prospectivity, as the Court discussed it in Johnson, is particularly suited for application in this case. A majority of our court has agreed that Almeida-Sanchez as applied to fixed checkpoints enunciates a new rule of criminal procedure. The sole purpose for the new rule is to deter future police misconduct. United States v. Calandra, 414 U.S. 338, 94 S.Ct. 613, 38 L.Ed.2d 561 (1974). Given this conclusion, the new rule can reasonably be applied only to convictions based upon searches conducted after the date of the Almeida-Sanchez decision. Freeing Bowen from his conviction will not deter future police misconduct.

If the new constitutional rule is designed to correct a defect that involves the court's ability to determine whether a defendant is guilty or not, it obviously has to be fully retroactive. But here, the new rule has no such effect. Bowen, like Almeida-Sanchez, is a convicted felon-- caught red-handed. The question is not whether they violated the law, but rather whether they should be freed because the police detected their violations through unconstitutional procedures. We are applying the Almeida-Sanchez rule to searches conducted at fixed checkpoints only to deter these unconstitutional procedures in the future. Our reason for applying the rule demonstrates that it should be applied prospectively only. Almeida-Sanchez receives the benefit of the new rule because of the Court's apparent belief that to do otherwise would put the Court in the position of providing advisory opinions.12 That policy is not enhanced or furthered by applying the rule to other cases on appeal. There is no compelling reason which requires that we overturn many convictions merely because the Supreme Court mandates that one must be overturned.

We conclude that any application of Almeida-Sanchez to fixed checkpoints would be a new rule and under the Stovall test, it would have to be applied prospectively to searches subsequent to June 21, 1973. Therefore, Almeida-Sanchez can be of no assistance to Bowen.

Affirmed.

Circuit Judges CHAMBERS, KOELSCH, EUGENE A. WRIGHT, TRASK, CHOY and SNEED concur in this majority opinion (Part II).

(Part I)

I respectfully dissent to Part I of the opinion pertaining to the applicability of Almeida-Sanchez to searches conducted at fixed checkpoints.

With one fell swoop, the majority in Part I hews down a law enforcement procedure used for 44 years to curtail the ever-increasing tidal wave of illegal aliens. The use of fixed checkpoints has been neither secret nor clandestine. The procedure has come before our court on numerous occasions1 with no hint that the practice was constitutionally infirm. For us to reverse ourselves at this late date requires clear and convincing reasons. See Smith v. United States, 273 F.2d 462, 465 (10th Cir. 1959), cert. denied, 363 U.S. 846, 80 S.Ct. 1619, 4 L.Ed.2d 1729 (1960). I fail to see them in the majority decision. The only apparent change is the opinion, or better said opinions, in Almeida-Sanchez v. United States, 413 U.S. 266, 93 S.Ct. 2535, 37 L.Ed.2d 596 (1973). A careful analysis, therefore, is required to see if that case dictates our reversal of a long-standing and accepted police practice.

The majority concedes that Almeida-Sanchez is a case involving stops and searches by roving border patrol officers. But by some mystic bridging, it holds that Almeida-Sanchez also requires that we outlaw searches at fixed checkpoints. The Opinion of the Court was delivered by Justice Stewart and concurred in by Justices Douglas, Brennan and Marshall. That opinion becomes the majority only with the added vote of Justice Powell. Justice Powell wrote:

While I join the opinion of the Court, which sufficiently establishes that none of our Fourth Amendment decisions supports the search conducted in this case, I add this concurring opinion to elaborate on my views as to the meaning of the Fourth Amendment in this context.'

413 U.S. at 275. Justice Powell's concurrence, therefore, adds a fifth Justice, and thus a majority, only to that part of the Justice Stewart opinion which invalidates the type of search 'conducted in (that) case'-- a roving patrol search.2

This critical distinction is brought into even closer focus by Justice Powell's demarcation of the four areas where searches typically occur:

(The search of Almeida-Sanchez' automobile) was not a border search, nor can it fairly be said to have been a search conducted at the 'functional equivalent' of the border. Nor does this case involve the constitutional propriety of searches at permanent or temporary check points removed from the border or its functional equivalent.

413 U.S. at 275-276. Thus, one could infer from his statement that searches can constitutionally occur at (1) the border, (2) functional equivalents of the border, (3) permanent checkpoints and (4) temporary checkpoints. He emphasized that the search in question did not occur in any of the four categories. His concurrence, therefore, cannot be said to give any weight to projecting Almeida-Sanchez to cover searches for aliens at fixed checkpoints. He specifically and emphatically limited his concurrence to answering the question of 'whether and under what circumstances the Border Patrol may lawfully conduct roving searches of automobiles in areas not far removed from the border for the purpose of apprehending aliens illegally entering or in the country.' 413 U.S. at 276.

Our brothers of the Tenth Circuit were faced with a question similar to that presented in this case in United States v. Bowman, 487 F.2d 1229 (10th Cir. 1973), which involved a stop at a fixed checkpoint located a few miles north of Truth or Consequences, New Mexico. Although they arrived at their conclusion by an alternate route, they discussed the effect of Almeida-Sanchez on the validity of the initial stopping of the defendant's vehicle and held: 'We therefore do not read the (Almeida-Sanchez) decision as challenging the right of immigration officials to make routine inquiries as to an individual's nationality.'3 The Tenth Circuit apparently has recognized that Almeida-Sanchez does not resolve the validity of immigration stops and searches at fixed checkpoints.

Ignoring the significance of Justice Powell's limited concurrence, the majority relies basically upon the language quoted by Justice Stewart from Carroll v. United States, 267 U.S. 132, 45 S.Ct. 280, 69 L.Ed. 543 (1925), and contends that that language demonstrates that Almeida-Sanchez also controls the question of searches at fixed checkpoints. However, here the majority falls into the trap of failing to realize that Justice Powell's concurrence does not necessarily make the language in the Justice Stewart opinion a pronouncement by a majority of the Court.

Because of the close fragmented vote and because Justice Stewart's opinion must be limited in application to roving patrols, great insight can be secured from the dissenting opinion of Justice White in which the Chief Justice and Justices Blackmun and Rehnquist concurred. Justice White, after noting that the Court in Carroll v. United States, 267 U.S. at 154, recognized that neither a warrant nor probable cause is required to stop and search at the borders, stated this right was undisputed in Almeida-Sanchez. He also, with the concurrence of the three other Justices, concluded: 'Neither, apparently, is it disputed that warrantless searches for aliens without probable cause may be made at fixed checkpoints away from the border.' 413 U.S. at 288. Thus, four Justices have specifically taken a position diametrically opposed to that adopted by the majority in this case. As Justice Powell has limited his opinion to the facts of the roving search, we are left to speculate as to his approach to the fixed checkpoint issue.4 Suffice it to say that based on the Almeida-Sanchez opinions, four Justices have concluded that searches for aliens at fixed checkpoints are not constitutionally infirm and that there is no majority consensus to the contrary. I, therefore, fail to see how the majority in this case can draw comfort from the Almeida-Sanchez decision.

After concluding that the search of Bowen's vehicle could be justified if it were conducted at a 'functional equivalent' of the border, the majority holds that as a matter of fact, this checkpoint is not a functional equivalent. Justice Stewart acknowledges in Almeida-Sanchez that border searches may be conducted at the functional equivalents of the border and gives us two examples. 413 U.S. at 272-273. But the examples give us precious little by way of definition. The majority attempts to expand that definition by referring to border extension cases where the vehicle involved has been under surveillance since it crossed the border, or where it is reasonably certain that persons had just illegally entered, or goods were just illegally smuggled into the United States. These cases are far different from checkpoint cases. They compare like apples and oranges. The only real assistance comes from the two examples in Justice Stewart's opinion. Yet, with this little guidance, the majority makes a factual finding that the checkpoint in this case was not a functional equivalent. How is this done? The majority merely states, based on 'the record.' The record in this case, however, only tells us where the fixed checkpoint is, how far it is from the border, that several highways intersect Highway 86 before the checkpoint, that there are several cities along the Highway before the checkpoint, that Bowen was stopped to see if any aliens were aboard his vehicle and that during the stop, contraband was discovered. There is no evidence as to the amount of traffic that comes through the checkpoint from the border area, the number of illegal aliens arrested at the checkpoint, or other information that would seem to bear on this question. The imprudence of deciding this issue on an incomplete record looms ever larger in view of the fact that a district court, after taking extensive testimony, concluded that the Route 86 checkpoint was a functional equivalent of the border. United States v. Baca, 368 F.Supp. 398 (S.D.Cal.1973). We would be far better advised to test the majority's theories on a factual record such as that developed in Baca rather than to make factual assumptions which may or may not be correct.5

This leads to the real question: What test does the majority apply to determine whether a fixed checkpoint meets the strictures of the Fourth Amendment?

It says that the government cannot justify searches made at fixed checkpoints without a warrant or probable cause on the basis of 8 U.S.C. 1357(a)(3) and 8 C.F.R. 287.1(a)(2) because the statute and regulation were vitiated in Almeida-Sanchez. But, once again, the majority fails to realize that the Justice Powell concurrence limits the Opinion of the Court so that it cannot be construed to apply to fixed checkpoints. But even if one ignores the significance of Justice Powell's concurrence, Justice Stewart did not state that 8 U.S.C. 1357(a) and the regulations issued pursuant to it were unconstitutional on their face, but merely that the statute could not vindicate a search that is otherwise in violation of the Fourth Amendment. Consequently, when the majority invalidates searches at fixed checkpoints simply because they too are conducted pursuant to 8 U.S.C. 1357(a) without independent consideration of whether such searches are reasonable under the Fourth Amendment, they bridge a gap with oversight rather than sound legal reasoning.

After finding as a fact that the checkpoint on Highway 86 was not a functional equivalent (in spite of there being no record upon which such a conclusion could be based) and after holding that the statutory power has been dissolved by Almeida-Sanchez (when there was no majority of the Supreme Court so holding), the majority concludes that the Fourth Amendment was violated because (1) 'the search was not the functional equivalent of a border search' and (2) it 'was authorized neither by warrant nor by probable cause.'6 The majority's test is incomplete. Four Supreme Court Justices have concluded that warrantless searches for aliens without probable cause may be made at fixed checkpoints. Their test was one of balancing the appropriate interests to determine whether the search was reasonable. Such an approach does not preclude Fourth Amendment analysis, but applies it without requiring that the search satisfy artificial criteria. It focuses upon the balancing protections afforded by the Fourth Amendment, as reiterated by Justice White, dissenting in Almeida-Sanchez: 'The Amendment's overriding prohibition is . . . against 'unreasonable' searches and seizures . . ..' 413 U.S. at 287-288. See Cady v. Dombrowski, 413 U.S. 433, 439, 93 S.Ct. 2523, 37 L.Ed.2d 706 (1973). Whether or not a search is reasonable 'depends upon the facts and circumstances of each case and . . . searches of cars that are constantly movable may make the search of a car without a warrant a reasonable one although the results might be the opposite in a search of a home, store, or other fixed piece of property.' Cady v. Dombrowski, 413 U.S. at 440, quoting Cooper v. California, 386 U.S. 58, 59, 87 S.Ct. 788, 17 L.Ed.2d 730 (1967).

In addition, we should not overlook the fact that the law enforcement personnel were operating pursuant to a statute. In such a case, there is even more reason to apply the broad test of reasonableness. See United States v. Biswell, 406 U.S. 311, 92 S.Ct. 1593, 32 L.Ed.2d 87 (1972); Colonnade Catering Corp. v. United States, 397 U.S. 72, 90 S.Ct. 774, 25 L.Ed.2d 60 (1970).

Certain types of searches have been constitutionally approved as reasonable although they have been made without probable cause or a warrant. A careful analysis of areas where such searches have been upheld as reasonable within the meaning of the Fourth Amendment demonstrates persuasive reasons for the same approach in testing the constitutionality of stops and searches at fixed checkpoints. For example, in Terry v. Ohio, 392 U.S. 1, 88 S.Ct. 1868, 20 L.Ed.2d 889 (1968), the Court concluded that such a search was reasonable after applying a test that balanced the interests of the individual in being free from invasions to his personal privacy against the interests of society in assuring the safety of its law enforcement officers. Under the circumstances of Terry, a pat-down without a warrant or 'probable cause' was not considered unreasonable. Similarly in Camara v. Municipal Court, 387 U.S. 523, 87 S.Ct. 1727, 18 L.Ed.2d 930 (1967), the Court upheld a non-objected-to7 search of all structures in a certain area to enforce a building code in spite of the nonexistence of probable cause or a warrant. Once again, the balancing test of reasonableness was applied to measure 'the need to search against the invasion which the search entails.' Id. at 537.

Stops and searches similar to the one objected to by Bowen occur not infrequently and, in many instances, are necessary facets of our way of life. The complexity of our society requires such limited intrusions in order to protect the rights of the majority.8 In these specific areas, as long as the stops and searches are limited in their scope to a proper objective and are not unreasonable, they satisfy the Fourth Amendment.

For example, all interstate motor carriers may be stopped for safety or other regulatory inspections. 49 U.S.C. 304 and 49 C.F.R. 396.5. Certainly, if contraband is discovered during such a search it could be used as evidence of the commission of a crime. Motor carriers are also stopped and weighed, which can be considered a limited search, as they enter a state. Such stops and searches have been upheld as legal. Commonwealth v. Abell, 275 Ky. 802, 122 S.W.2d 757 (Ky.App.1938). Similarly, probable cause is not necessary to stop a vehicle to check for a valid driver's license and contraband found incident to such a stop is not the subject of an illegal search.9 United States v. Croft, 429 F.2d 884 (10th Cir. 1970); Lipton v. United States, 348 F.2d 591 (9th Cir. 1965). Another example is the routine searches of baggage and other personal effects of persons entering the mainland from Hawaii for infectious plants which has also been upheld as reasonable within the Fourth Amendment. United States v. Schafer, 461 F.2d 856 (9th Cir. 1972).

Perhaps the most obvious example is the limited search that all passengers on airline flights are subjected to before they are even allowed in the boarding areas. At least at one time, if the would-be passenger met additional criteria (supposedly objective but which were secretly held by officers and airline employees), he could be subjected to an even more detailed search. We have repeatedly held that evidence of a crime found in the course of such a search may not be suppressed on the theory that the search violates the Fourth Amendment. E.g., United States v. Doran, 482 F.2d 929 (1973); United States v. Davis, 482 F.2d 893 (1973). When the interest of the public in being protected against the dangers of a highjacking were weighed against the intrusion, we concluded that such searches were reasonable.10 As we stated in United States v. Davis, 482 F.2d at 910:

To pass constitutional muster, an administrative search must meet the Fourth Amendment's standard of reasonableness. 'Unfortunately, there can be no ready test for determining reasonableness other than by balancing the need to search against the invasion which the search entails.' Camara v. Municipal Court, supra, 387 U.S. at 536-537, 87 S.Ct. at 1735.

I would not restrict the constitutional test in this fact situation as does the majority but would determine whether the search at this fixed checkpoint was unreasonable under the Fourth Amendment. The majority holds that:

The opinion in Almeida-Sanchez, delivered by Mr. Justice Stewart, leaves little doubt that traditional Fourth Amendment standards apply to fixed-checkpoint searches as well as to roving-patrol searches.

But there is a majority holding only on the issue of roving patrols. There is a vast difference between a red-light, midnight stop on a lonely road and approaching a well-lighted, fixed checkpoint with warning signs and uniformed men plainly visible.11 The majority's assumption that five of the Justices have or would apply the same standard in both is one I cannot accept.

Under the correct test, whether the search at this fixed checkpoint was unreasonable should be determined in the first instance by the trial court after hearing all of the evidence. Such a determination is primarily factual. In applying this test the trial court would balance the rights of the individual vehicle driver against the interests of all the people of the country in stemming an avalanche of persons illegally crossing our borders.12 Both Justices Powell13 and White14 refer to the Herculean challenges faced by those directed to prevent illegal entries. Before we direct the Dutch boy to remove his finger from the dike, we owe it to the American people at least to balance their interests against the interests of the individual in being free from this limited intrusion.

While no one can give an accurate count, it has been estimated that there are approximately 800,000 to over 1 million illegal aliens in our country; approximately 85% Of these are citizens of Mexico. United States v. Baca, 368 F.Supp. at 402. The major problem is the abortive attempt to guard 2,000 miles of border with Mexico from the Gulf of Mexico to the Pacific coast.

In addition to aliens entering illegally, 91 million aliens entered the United States legally during the fiscal year 1972, without over 39 million of these entering directly into Southern California. Id. at 404. A large percentage of these visitors enter with temporary border passes, restricting them to a seventy-two hour stay and to travel within twenty-five miles from the border. Id. at 404; see 8 C.F.R. 212.6. To enforce these restrictions, some type of investigation away from the border is required. The majority suggests that this flow of illegal aliens can be sufficiently regulated by alternative methods. But I suggest that we do not have sufficient information in this record to make that determination. Indications are that anything short of an Iron Curtain type border patrol would be ineffective in curtailing the number of illegal entries. United States v. Baca, 368 F.Supp. at 405. Further, even stopping the illegal flow across the border would not halt those legally entering but illegally travelling more than twenty-five miles from the border. Therefore, while I agree with the majority that alternative methods of enforcement of the law should be included in the balancing approach, one district court found:

The evidence presented before this court clearly established that there is no reasonable or effective alternative method of detection and apprehension available to the Border Patrol in the absence of the checkpoints, for even a geometric increase in its personnel or line watch would not leave any control over those admitted as temporary visitors from Mexico.

Id. at 408. Certainly from the record before us, we cannot make a contrary conclusion.

If we are to test the search in this case as to whether it is unreasonable under the Fourth Amendment, it would be necessary to remand the matter for a factual determination. The findings already made by one district court pertaining to the checkpoint where Bowen was stopped indicate that such would not be a fruitless exercise.15 With an adequate factual record, we could proceed with the delicate constitutional balancing necessary to determine whether the stop and search were unreasonable.

Circuit Judges CHAMBERS, KOELSCH, EUGENE A. WRIGHT, CHOY and SNEED concur in this dissenting opinion to Part I.

(Part II)

ALFRED T. GOODWIN, Circuit Judge (concurring and dissenting):

Believing that the legal history of roving patrols is fundamentally different from the legal history of fixed-checkpoint searches, I concur in that part of Judge Wallace's opinion in Part II which limits the holding of Part I to searches conducted after June 23, 1973. While a roving-patrol search, supported by neither warrant nor probable cause, was not upheld by this circuit until 1970 (United States v. Miranda, 426 F.2d 283 (9th Cir. 1970)), a mere two years before certiorari was granted in Almeida-Sanchez (406 U.S. 944, 92 S.Ct. 2050, 32 L.Ed.2d 331 (1972)), fixed-checkpoint searches enjoyed judicial approval at least since 1963 (Fernandez v. United States, 321 F.2d 283 (9th Cir. 1963)) and apparent statutory authorization since 1946 (Act of Aug. 7, 1946, ch. 768, 60 Stat. 865.) Our rejection of fixed-checkpoint searches therefore 'marks a sharp break in the web of the law,' Milton v. Wainwright, 407 U.S. 371, 381-382 n. 2, 92 S.Ct. 2174, 2180, 33 L.Ed.2d 1 (1971) (dissenting opinion of Stewart, J.), sufficient to deny retroactive application.

My concurrence is qualified, however, by the belief that Bowen himself should be entitled to the fruits of his appeal. He fully preserved his objections to the search of his camper-truck by the border patrol. He took his case all the way to the Supreme Court, which reversed his conviction and remanded the case for reconsideration in light of Almeida-Sanchez. Now, we tell Bowen that he was right, that searches at fixed checkpoints, supported by neither a warrant nor probable cause, are unconstitutional. Yet, we also tell him that because of a judicially created rule of nonretroactivity, he cannot take advantage of a ruling which he has fought for two years to obtain. However, regardless of the nonretroactivity of our newly announced rule on fixed checkpoints, as the moving party in the case in which this new rule has been announced, Bowen is fully entitled to its protection. To prevent having our determination in Part I stand as mere dictum, as well as to comply with the 'case or controversy' requirement of Article III of the Constitution, we must allow Bowen to benefit from our constitutional determination. See Stovall v. Denno, 388 U.S. 293, 301, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967). See also Desist v. United States, 394 U.S. 244, 254-255 n. 24, 89 S.Ct. 1030, 22 L.Ed.2d 248 (1969).

Circuit Judges MERRILL, BROWNING and DUNIWAY concur in this concurring and dissenting opinion.

HUFSTEDLER, Circuit Judge, with whom Circuit Judge ELY concurs, concurring in part I and dissenting from part II, of the opinion of the Court:

I cannot concur in the opinions confining the Almeida-Sanchez rule to searches occurring after June 21, 1973, with or without an exception for Bowen himself, when the locale of the search was a fixed checkpoint that was not the functional equivalent of an international border.

My brothers and I agree that no retroactivity issue arises unless Almeida-Sanchez stated a new rule marking 'a sharp break in the web of the law.'1 We are also in agreement that Almeida-Sanchez announced no new Supreme Court rule and that it signaled no break at all in the constitutional principles stated by the Court.2 Injection of the retroactivity issue in the opinions of my brothers Wallace and Goodwin rests on two assumptions: (1) Congress, the Ninth Circuit, and the Immigration and Naturalization Service, prior to Almeida-Sanchez, had long agreed that automobile searches conducted by personnel of the Service at fixed checkpoints that were not functional equivalents of international boundaries should be treated as if they were such 'border searches,' thus eliminating the Fourth Amendment's requirements of a warrant and probable cause, and (2) for retroactivity purposes, the weavers of the constitutional law web are not limited to the members of the Supreme Court, but also include Congress, the lower federal courts, and the personnel of the Immigration and Naturalization Service. The first assumption is inaccurate as a matter of fact; the second is wrong as a matter of law.

My brothers correctly state that the Immigration and Naturalization Service began establishing fixed checkpoints some distance from international boundaries in 1929. The information before us about those checkpoints is fragmentary. We do not know how many there were, where they were located, or which, if any of them, qualified as functionally equivalent to international boundaries. We also lack information about the kinds of searches, if any, that were conducted at them. However, we do know that any vehicle searches that were undertaken at points other than international boundaries were without color of statutory authority until 1946 when Congress enacted the precursor of 8 U.S.C. 1357. (See Act of Aug. 7, 1946, ch. 768, 60 Stat. 865).3 Although Congress thus eventually authorized checkpoint searches by Immigration and Naturalization personnel, congressional attention never focused on the constitutional difficulties engendered by the movable border concept that was included in section 1357. (See, e.g., H.R.Rep.No.186, 79th Cong., 2d Sess., supra note 3.)

Fourth Amendment challenges to 'alien' searches conducted by immigration and customs officials at places other than international boundaries were raised in the lower federal courts. The judicial response, however, did not rest on a comprehensive, critical analysis of the differences between searches at international boundaries and those conducted some distance from the border;4 rather, the law concerning checkpoint and roving patrol searches developed on a case-by-case basis, following the practice so familiar in other Fourth Amendment contexts. The end product of the evolution in this circuit was that section 1357 became a justification for the initial stop of a vehicle and the preliminary interrogation of the occupants at either a fixed checkpoint or elsewhere within a reasonable distance of the border; but probable cause was still required to validate a warrantless search of the vehicle, at least if the search went beyond that reasonably related to the discovery of aliens.5 Although these 'alien' non-international boundary vehicle search cases were sometimes labeled 'border searches,' they were always a breed apart from searches actually conducted at international boundaries. The latter were treated as they always had been: no warrant and no probable cause were required to validate a thorough, fullscale search of the vehicle and its contents. The validity of warrantless, nonboundary 'border searches' conducted without probable cause, on the other hand, was limited both in scope and in object.

Of perhaps greater moment for our retroactivity analysis, an examination of Ninth Circuit cases reveals that fixed checkpoint searches fared neither better nor worse than searches conducted at temporary checkpoints or by roving patrols away from the border. In all of these cases we approved initial stops pursuant to section 1357 and limited the scope of subsequent search activity. (See note 5 supra.) The assumption of my brothers that full-scale international boundary-type vehicle searches conducted at fixed checkpoints located away from the border had received long-continued judicial approval is not supported by the facts; my brothers' conclusion that Almeida-Sanchez sharply broke with prior law from the lower federal courts concerning fixed checkpoint searches cannot be sustained.

Even if the Ninth Circuit for many years had consistently upheld these alien searches at fixed checkpoints, however, that fact would not be pertinent to deciding whether retroactivity is in issue.6 For this purpose, the only 'old' and 'new' law that is relevant is the constitutional law enunciated by the Supreme Court. (See Robinson v. Neil (1973), 409 U.S. 505, 510, 93 S.Ct. 876, 35 L.Ed.2d 29.) A contrary rule would impair or destroy the uniform application of constitutional law in the several circuits, increase manifold the certiorari burdens of the Supreme Court, and erode the power of the Supreme Court as the ultimate arbiter of constitutional law.

These disturbing consequences might not occur if all circuits simultaneously decided constitutional questions and always decided them the same way. But they do not. The question of the constitutionality of fixed checkpoint searches at issue in the case at bench provides an excellent illustration of the problems that can arise under my brothers' views. The issue arose at different times with different frequency in some circuits and in others did not arise at all. In the circuits where the issue was never decided, did Almeida-Sanchez state a 'new' rule? Is a single decision in one circuit upholding the search enough to create 'old' law with which Almeida-Sanchez broke? If one circuit had upheld such searches and another had struck them down, would Almeida-Sanchez be retroactive in the latter circuit and nonretroactive in the former?7 To prevent restrictions on the retroactivity of its decisions, must the Supreme Court take every new search and seizure issue to prevent 'old' law from accumulating? If the Court is in any respect bound by the constitutional law developed in the lower federal courts, how can it continue to be the final arbiter of constitutional issues?8

Even when the retroactivity concept is confined exclusively to law stated by the Supreme Court, difficult jurisprudential problems are generated. (See, e.g., Desist v. United States (1969) 394 U.S. 244, 256-269, 89 S.Ct. 1030, 22 L.Ed.2d 248 (Harlan, J., dissenting).) To extend the concept to embrace the law of the circuits would turn confusion into chaos.

CHAMBERS, Circuit Judge (concurring and dissenting):

I concur and dissent as indicated above.

I adhere to my original position that these Almeida-Sanchez cases should not have been taken en banc. See United States v. Bowen, 9 Cir., 485 F.2d 1388.