(dissenting):

I respectfully dissent. 1 To me, the manner in which the majority disposes of a highly significant constitutional issue is unacceptable.

In Sherriek v. Eyman, 389 F.2d 648 (9th Cir. 1968), we reviewed the Arizona search warrant issuance procedure that was here employed. I dissented from the majority opinion in that case because it held that certain unspecified oral “information,” given to the magistrate by the affiant police officer at the time the warrant was requested could and, in fact, did “cure” the inadequacy of the officer’s written affidavit. It was my opinion that the challenged procedure thwarted subsequent, independent, and adequate review of the basis for the issuance of a search warrant and was constitutionally infirm. 389 F.2d at 654 . I wrote:

I still strongly adhere to that position. I do not believe that the constitutional requirements imposed upon government officers seeking warrants, see, e. g., Aguilar v. Texas, 378 U.S. 108 , 84 S.Ct. 1509 , 12 L.Ed.2d 723 (1964), can be, in effect, overborne by the operation of state procedural law. That, however, is what the majority herein allows.

The shortcoming in the present case is not as egregious as it was in Sherriek , since, marking an improvement, there is some evidence of what the affiant officers told the magistrate. The problem, ignored by the majority, is, simply, that that evidence is confused and unreliable. The majority opinion may be ready to imply that a transcript was made of the proceedings conducted when the warrant was requested. That is not the fact. The only evidence of what transpired when the warrant was issued is to be found in the transcript of Boyer’s trial. During the trial, which was conducted in July, 1969, almost seven months after the search was authorized, the police officers and the magistrate testified. They related, as best they could, their memories as to the events of the January 3d proceeding. It is from this, alone, that the majority concludes, in effect, that there was a substantial basis for the issuance of the warrant.

The problem inherent in relying upon belated oral testimony to find that the existence of probable cause was demonstrated at an earlier hearing is obvious. Memories are blurred by the passage of time and by the wisdom gained through hindsight. Critical facts may be forgotten, and the possible initial uncertainty of the affiant may vanish when the search proves to be fruitful. Inadvertent additions to the remembered conversation are not unlikely. 2 Since we test the basis for search warrants in a common sense and realistic fashion, see United States v. Ventresea, 380 U.S. 102 , 85 S.Ct. 741 , 13 L.Ed.2d 684 (1965), many of the errors which unwittingly appear in this type of testimony are harmless and we could, therefore, accept Arizona’s procedure as being without prejudice to the defendants in most cases. But prejudicial errors can also occur, and all errors could be avoided by requiring the recordation of oral testimony received by the magistrate to supplement an affidavit for a warrant. I believe that the risk of error we are asked to ignore in this case places an in *808 tolerable restriction upon our power effectively to review a magistrate’s decision as to probable cause.

Here, the transcript, as presented to us, 3 proves the reality of the dangers that I have postulated. The police officers who had sought and obtained the warrant testified at length, with remarkable recall and certainty. As they recounted the proceedings, the information orally supplied by them to the magistrate clearly met constitutional standards. See United States v. Harris, 403 U.S. 573 , 91 S.Ct. 2075 , 29 L.Ed.2d 723 (1971); Spinelli v. United States, 393 U.S. 410 , 89 S.Ct. 584 , 21 L.Ed.2d 637 (1969); Aguilar, supra. In contrast, the testimony of the magistrate was less complete. Much of his testimony was weakened by qualification, such as “Well, it’s pretty difficult [to recall], but”, “As I recall”, and “It’s a little difficult to recall, but”. To me, it appears that he strove to testify honestly; hence, his recollection of the unreported proceedings was, understandably, less than perfect.

The most significant aspect of the magistrate’s difficulties in recalling the incident is the fact that, on the record before us, as the magistrate remembered the oral information supplied to him by the police, it was insufficient under Harris, Spinelli, and Aguilar . Those cases hold that when an application for a warrant is based upon an informer’s tip, the existence of probable cause depends upon satisfaction of a two-part test. As the majority notes, the police must demonstrate to the magistrate

Aguilar, supra, 378 U.S. at 114 , 84 S.Ct. at 1514 . 4 Our limited record of the magistrate’s testimony includes ample information showing that the informant was reliable, but it contains nothing which shows that the police revealed how or why the informant concluded “that the narcotics were where he claimed they were.” Indeed, there is an indication that the police failed to do so:

From this, it appears that, contrary to the majority’s interpretation of the record, the magistrate was not advised as to how the informer acquired his information. Nor was he advised that the informer was a close personal friend of Boyer who had detailed factual information which would lead the magistrate to “know that he [was] relying on something more substantial than a casual rumor circulating in the underworld or an accusation based merely on [the suspect’s] general reputation.” 5 If I am correct, the Aguilar test was not met in this case, the warrant should not have been issued, and the search was illegal.

The significant discrepancies between the proffered versions of the January 3d proceedings clearly mark the evil promoted by the Arizona procedure. In some cases, as here, it will be obvious that there is confusion as to what ac *809 tually transpired when the warrant was requested. In other cases, confusion may be obscured because of the absence of obvious discrepancies in the testimony of different people. But in either case, there exists a substantial possibility that a reviewing court will be induced to make a wrong decision. Such risks can be readily avoided without impairing law enforcement and with full protection of Fourth Amendment rights. I would hold that an affidavit in support of an application for a search warrant must, in and of itself, be sufficient, or that any sworn oral testimony presented to supplement a defective affidavit must be accurately transcribed at the time of its original presentation. Accordingly, I would reverse.

. If I could find adequate reliable evidence of the facts outlined in the majority opinion, I would readily concur with that opinion. The procedures followed in this case, however, trouble me greatly.

. Such possible additions are of critical importance, for “it is elementary that in passing on the validity of a warrant the reviewing court may consider only information brought to the magistrate’s attention.” (Emphasis in original.) Aguilar, supra, at 109, n. 1, 84 S.Ct. at 1511 , citing Giordenello v. United States, 357 U.S. 480, 486 , 78 S.Ct. 1245, 1250 , 2 L.Ed.2d 1503, 1509-1510 (1958).

. The record before this court of the testimony offered by the police officers and the magistrate is incomplete. It is, of course, the burden of the state to prove that there was sufficient evidence offered when the warrant was requested.

. This test is explicitly recognized in Harris and Spinelli .

. This is the test of “underlying circumstances” articulated in Spinelli, supra, 393 U.S. at 416 , 89 S.Ct. at 589 , 21 L.Ed.2d at 643-644 (1969). It is irrelevant that the police could have supplied such information. They must have actually done so. See note 2, supra.