dissenting.
The government argues that the combination of wholly ambiguous testimony from a handwriting expert and equivocal testimony from a witness receiving favorable treatment from the government is sufficient to support the conviction of Altigraci Rosario for passing a United States Treasury cheek. The majority accepts this argument. I do not; hence, I dissent.
To convict Rosario of check forgery under 18 U.S.C. § 510(a)(2), the government was required to prove four elements beyond a reasonable doubt: (1) that the check was a U.S. Treasury check; (2) that the check bore a forged or falsely made endorsement; (3) that Rosario passed the check with intent to defraud; and (4) that Rosario acted knowingly and willfully. There was no direct evidence adduced at trial to satisfy the government’s burden on elements (2), (3) and (4). Recognizing this, the government nonetheless asks us to cobble together a series of inferences to support the jury’s verdict. It argues that, taken collectively, the testimony of Taylor, the handwriting expert, and Rios, the man who negotiated the stolen check, are sufficient to permit the jury to infer that Rosario forged the check. Building on this inference, it then claims that the jury could draw the further inferences that Rosario possessed the requisite knowledge, willfulness and intent to defraud necessary to satisfy the remaining elements of the charged offense. In my view, these “inferences” do no more than permit the jury to speculate that Rosario is guilty, especially in light of the weak testimony from which these inferences are drawn.
Jeffrey Taylor, the government’s handwriting “expert,” could only testify that Rosario “probably” signed the name “Ana Andrade” to the back of the Andrades’ check. The trial record shows, however, that Taylor’s testimony was even more ambiguous. Indeed, under cross-examination Taylor conceded that there were a number of “irreconcilable differences” between the Ana Andrade signature on the check and Rosario’s sample signature. App. at 35A. Moreover, Taylor candidly admitted that there was “some doubt” in his mind as to whether Rosario signed Ana Andrade’s name on the check. App. at 35A-36A. Significantly, Taylor also acknowledged on direct examination that he “found no evidence that [Rosario] wrote the remaining signature [Angel Andrade’s] on that check.” App. at 32A. Taylor’s concessions make his already equivocal conclusion that Rosario “probably” forged Ana Andrade’s name on the check even less reliable. I would conclude that inferences drawn from such clearly ambiguous testimony cannot possibly satisfy the government’s burden of establishing beyond a reasonable doubt that Rosario forged Ana Andrade’s signature on the check.
Recognizing the inherent weakness of Taylor’s vague opinion, the government would have us rely on the testimony of Rios for support that Rosario forged the check. Rios’s testimony, it argues, establishes that Rosario both possessed and had the opportunity to forge the check, thereby allowing the jury to infer that Rosario did, in fact, forge Ana Andrade’s signature on the check. By presenting evidence that Rosario possessed the cheek and had the opportunity to sign it, the government contends that it provided validation for Taylor’s equivocal opinion that Rosario probably forged the check. In support of its argument, the government relies primarily on two cases where courts affirmed forgery convictions based in part on testimo*166ny from a handwriting expert indicating that the defendant had “probably” forged the stolen check. See United States v. Richardson, 755 F.2d 685 (8th Cir.1985) (per curiam); United States v. Rivamonte, 666 F.2d 515 (11th Cir.1982) (per curiam).
In my view, however, reliance on Richardson and Rivamonte is imprudent for a number of reasons. First, notwithstanding the assertion that the government offered only “slightly” more circumstantial evidence in Richardson and Rivamonte than that adduced here, Maj. Opinion at 164, the records in those cases demonstrate that there was ample evidence tending to establish all elements of those check forgery convictions.
For example, in Richardson, the court affirmed a check forgery conviction where the handwriting expert’s testimony was complemented by evidence showing that Richardson had a key to the home where the check was stolen, Richardson’s fingerprints were found on the stolen check, a stolen deposit slip was used to cash the check, and Richardson had made a deposit in the exact same amount as the stolen check during the time period in which the stolen check was cashed. 755 F.2d at 686.
Similarly, in Rivamonte, the court affirmed a check forgery conviction where the expert’s opinion was complemented by evidence showing that Rivamonte’s fingerprints and palmprints were found on the check, the defendant’s account number was written on the back of the check, the payee’s names were written on Rivamonte’s pre-encoded deposit slip, and a deposit was made in the defendant’s account on the same day that the stolen check was negotiated. 666 F.2d at 516-17.
In each case, the government proffered strong circumstantial evidence specifically related to the respective defendants’ possession of the stolen checks, their intent to defraud and their states of mind. Such was not the case here, where the government, lacking sufficient evidence to establish any of these elements beyond a reasonable doubt, was forced to ask the jury to speculate that Rosario forged the check, passed the check with intent to defraud, and acted with requisite knowledge and willfulness.
I do not believe we can contort Richardson and Rivamonte to support the proposition that testimony from a handwriting expert indicating that a defendant “probably” forged a stolen check in conjunction with evidence showing possession of the stolen check by the defendant constitutes sufficient evidence to affirm a conviction under 18 U.S.C. § 510(a)(2). Simply stated, there is no such baseline position established in the case law. Instead, Richardson and Rivamonte suggest that an “expert” opinion that the defendant probably forged the check, coupled with sufficient additional circumstantial evidence demonstrating possession, willfulness, knowledge and intent to defraud, is necessary before a conviction will be affirmed.
United States v. Hall, 632 F.2d 500 (5th Cir.1980), is not to the contrary. In Hall, the court held that once forgery is conclusively proven, inferences of fact regarding possession, intent and knowledge can be permissibly drawn by the government. Id. at 502. The handwriting expert in Hall, however, provided an unequivocal opinion that the defendant had forged the payee’s name on the stolen cheek, thereby providing the government with conclusive factual proof of the forgery element of the offense from which inferences tending to establish the other elements of the offense could be drawn. Id. Here, in contrast, the government has offered only ambiguous, inconclusive testimony regarding the forgery element of the offense. As such, there is no conclusively proven fact of forgery from which the government could draw inferences tending to establish the other elements of the offense of conviction.
My interpretation of the case law is supported by the post -Rivamonte decision in United States v. Henderson, 693 F.2d 1028 (11th Cir.1982), which, in my view, does not bolster the government’s argument. In Henderson, the court reversed a check forgery conviction based on ambiguous handwriting testimony and circumstantial evidence tending to show that the defendant’s wife had cashed the stolen check. In reaching its decision, the court reasoned as follows:
*167Although it is apparent that someone endorsed Mr. Moore’s signature on the back of the treasury check, the evidence was not sufficient for a fair jury to conclude beyond a reasonable doubt that Mr. Henderson was the endorser. The evidence, because it was circumstantial required that the jury draw an inference that because Ms. Henderson used the defendant’s ear to cash the check, and because Ms. Henderson did cash the check, the defendant must have signed the check. This simply does not follow. It is unreasonable to infer Mr. Henderson’s guilt based upon the actions of his wife. Yet, it is apparent from the evidence that there was little else upon which to base a conviction---- Although circumstantial evidence is testimony to the surrounding facts and circumstances of the point at issue, they must at some point connect, to allow the trier of fact to draw the inference that the fact asserted is true.
Id. at 1031 (internal citation omitted). The court then proceeded to distinguish Rivamonte on the basis of the strength of the additional evidence offered by the government in that case. As the Henderson court concluded: “In the present ease, the additional evidence, together with the handwriting expert’s ‘probable’ testimony, is not sufficient.” 693 F.2d at 1032. Significantly, there is nothing in the Henderson decision to suggest that the court viewed the failure of the government to produce evidence showing that Mr. Henderson possessed the stolen cheek as determinative of the sufficiency of the evidence. Rather, the Henderson court reviewed the proffered evidence in its entirety and determined that there was insufficient evidence supplementing the ambiguous handwriting testimony to permit a reasonable jury to conclude beyond a reasonable doubt that Mr. Henderson was guilty of the offense of conviction.1
Notwithstanding the absence of any legal precedent for its conclusion that ambiguous handwriting evidence coupled with evidence of possession constitutes sufficient evidence to affirm a conviction under § 510(a), the government speciously reasons that Rosario’s conviction was proper because Rios’s testimony that Rosario possessed the check provided the same corroboration for the handwriting expert’s testimony that the fingerprint evidence in Rivamonte and Richardson did. What this bit of forensic gymnastics neglects to explain, however, is that the government’s fingerprint expert was unable to identify any finger or palm prints belonging to Rosario on the Andrade check. App. at 42A-43A. Thus, the government was forced to rely on Rios’s testimony as the “equivalent” of fingerprint evidence precisely because there was no fingerprint evidence available to support the conclusion that Rosario forged Ana Andrade’s name on the back of the stolen check. Rather than lend credibility to the ambiguous handwriting testimony offered in this case, the government’s reliance on Rios’s testimony highlights the dearth of evidence offered by the government to meet its burden of proof. Simply stated, aside from Rios’s testimony the government failed to adduce any additional evidence to validate Taylor’s equivocal conclusion that Rosario signed the stolen cheek. Lacking further additional evidence like that offered in the Rivamonte and Richardson cases (e.g., fingerprints, palmprints, pre-coded deposit slips), I fail to understand how Rios’s testimony could possibly transform Taylor’s ambiguous conclusion into factual proof sufficient to establish Rosario’s guilt beyond a reasonable doubt.
Finally, I am concerned because parts of Rios’s testimony directly contradict inferences that the jury was supposed to have drawn from Rios’s testimony. For instance, on cross-examination Rios testified that Rosario did not know that the check was stolen. App. at 52A. Such testimony clearly undercuts the idea that the jury could infer that Rosario had the requisite knowledge and intent to defraud necessary to support a conviction under § 510(a)(2). Moreover, it also puts the majority in the awkward position of relying on Rios’s testimony in order to bolster the inferences that Rosario possessed *168and forged the stolen check, but ignoring Rios’s testimony in order to draw the inferences that Rosario had the requisite knowledge and state of mind necessary to support her conviction. Such inconsistencies further reinforce my conclusion that the evidence proffered in this ease permitted the jury to do little more than speculate as to Rosario’s guilt.
In summary, I believe that the evidence adduced by the government at trial falls far below the horizon of certainty we require in criminal prosecutions and is not sufficient to convict Rosario beyond a reasonable doubt. Handwriting analysis is at best an inexact science, and at worst mere speculation itself. See, e.g., D. Michael Risinger et al., Exorcism of Ignorance as a Proxy for Rational Knowledge: The Lessons of Handwriting Identification “Expertise”, 137 U. Pa. L.Rev. 731, 739 (1989) (reporting that “[f|rom the perspective of published empirical verification, handwriting identification expertise is almost nonexistent”). As such, I do not believe that wholly ambiguous testimony from a handwriting “expert” and selected testimony from a witness receiving favorable treatment from the government can satisfy the government’s burden of proof. Accordingly, I would reverse Rosario’s conviction.
. The majority correctly states that in Henderson the government offered no evidence that Mr. Henderson had ever possessed the stolen check. Maj. Opinion at 164. I note, however, that the government similarly failed to offer any evidence specifically relating to Mr. Henderson’s intent to defraud, knowledge or state of mind.