(concurring specially):
It has been long thought that a federal rule exists, to wit: That a defendant can be convicted solely on the uncorroborated testimony of an accomplice witness. This “rule” supposedly grew out of the cases of Holmgren v. U. S., 217 U.S. 509, 30 S.Ct. 588, 54 L.Ed. 861 (1910), and later Caminetti v. U. S., 242 U.S. 470, 37 S.Ct. 192, 61 L.Ed. 442 (1917). In Caminetti the court said, at page 495, 37 S.Ct. at page 198, it is:
“. . . the better practice for courts to caution juries against too much reliance upon the testimony of accomplices, and to require corroborating testimony before giving credence to such evidence. While this is so, there is no absolute rule of law preventing convictions on the testimony of accomplices if juries believe them.”
It is important to note that the key word “uncorroborated” was omitted from the last sentence in the above quote. That omission was not unintentional. Although the testimony of an accomplice witness in the case was most damaging to the defendant, there was indeed some important corroboration. The defendant took the stand in his own behalf, but declined to answer questions detrimental to his case or put forth defenses to them. The corroboration was the inference to be drawn from the defendant’s silence concerning incriminating matters. Therefore, the court in Caminetti could not have been concerned with the issue of uncorroborated testimony of an accomplice as the only convicting evidence. The court there gave the real test when it said, at pages 495-496, 37 S.Ct. at page 198:
“This court does not weigh the evidence in a proceeding of this character, and it is enough to say that there was substantial testimony to support the verdict rendered in the trial courts.” (emphasis supplied)
*463In Caminetti Mr. Justice Day, in affirming the Circuit Court of Appeals decision, said that accomplice instructions “in the form made should not have been given.” This is not the same thing as saying that proper instruction ought not to be given, nor that a conviction cannot be reversed where there is no corroborations and especially where we have the sort of record so ably delineated by Judge Duniway in his opinion. On the contrary, the Supreme Court has indicated rather clearly what is the better practice, Holmgren v. U. S., supra.
If there is “no absolute rule” preventing convictions — then, in the opinion of this writer, there is “no absolute rule” requiring convictions on uncorroborated testimony nor preventing an appellate court from reversing a conviction where there is only the testimony of an accomplice and especially where that testimony is shattered and shredded by contradictions and other asserted errors appear. Lyda v. United States, 9 Cir., 1963, 321 F.2d 788, 795. This is not to say nor suggest that the appellate court should weigh the testimony of the accomplice. Only that it should search the record to determine in a situation such as this whether (1) there is corroboration as to some material points, especially if the testimony of the accomplice is incredible, unsubstantial or seriously contradicted, (2) correct accomplice instructions have been requested or given, and (3) from the standpoint of other asserted errors, that in the whole overview of the case it is assured that there has not been a miscarriage of justice.
The salutary application of the rule (if it be the rule) that a conviction can be had in every case upon the uncorroborated testimony of an accomplice is dangerous in the extreme. It invites perjury and a lying contest. Inherently, the testimony of an accomplice is weak, suspect, and insidious. If “it is better practice for „ courts to caution juries against too much reliance upon the testimony of accomplices, and to require corroborating testimony” before giving credence thereto, then, in the opinion of this writer, such “better practice” commands that corroboration be the rule, especially as applied to this factual situation.
Later Ninth Circuit cases which are said to further the “rule” set forth in Caminetti fit into the same pattern. That is to say, there exists at least some form of corroboration in addition to the testimony of an accomplice witness. Therefore, if there is in fact such a federal rule, courts have only had to pay lip service to it. In Moody v. U. S., 376 F.2d 525 (1967), the defendant was convicted of illegal possession of heroin found in his car. The testimony of an accomplice clearly led to the defendant’s conviction, but nevertheless that testimony was supported by the fact that the heroin was found in the defendant’s car, a very significant corroborating fact.
Likewise, in the more recent ease of Suhl v. U. S., 390 F.2d 547 (1968), which the United States relies on heavily, there was a type of corroboration. There the defendant was convicted of mail fraud involving a check-kiting scheme. The accomplice witness changed his story between the two trials, as well as on cross-examination at the second trial, as to who signed the checks. The difference between the Suhl case and the one of concern now is that in Suhl the jury had the additional corroboration of the physical evidence of the checks with defendant’s signature on them.
In the ease at bar there is absolutely no form of corroboration, no physical evidence, no inferences to be drawn from the defendant’s testimony, only the testimony of an accomplice witness who had every reason to fabricate. Surely the dictum in the Caminetti case could not be extended to cover the facts of this case.
Under the circumstances of this factual situation, Haynes’ testimony, standing alone and without corroboration, should not be held to be sufficient to convict.
For these reasons and those expressed by Judge Duniway, with the exceptions *464noted, I concur in reversal and remand for a new trial.