UNITED STATES COURT OF APPEALS FOR THE D.C. CIRCUIT


UNITED STATES

v.

WILLIAMS, GREGORY


00-3003a

D.C. Cir. 2000


*	*	*


Randolph, Circuit Judge: The issue is whether the perjury  of a
prosecution expert witness, discovered after trial, entitled  Gregory
L. Williams to a new trial. The jury convicted  Williams of possession
with intent to distribute heroin, in  violation of 21 U.S.C. s
841(a)(1). When arrested, Williams  had 10.5 grams of heroin in 87
small plastic "baggies" in his  coat pocket, and was storing another
75.3 grams of heroin,  packaged in 638 small plastic "baggies," in an
automobile.  One government expert, a forensic chemist, testified
about  the tests he performed to establish that the material in the 


Another government witness, Detective Johnny St. Valen- tine-Brown,
answered a question about his qualifications,  stating that he had
been a narcotics expert for more than  twenty years and had served as
a senior narcotics policy  analyst in the Reagan and Bush
administrations. At the end  of his lengthy response, he added: "I am
also a Board- certified pharmacist. I receive, maintain compound and
dis- pense narcotic, as well as non-narcotic substances per pre-
scription." Without objection, the court accepted Brown as  an expert
in the "distribution and use of narcotics, the  packaging of narcotics
for street-level distribution, the man- ner in which narcotic dealers
distribute narcotic substances in  the District of Columbia, the price
for which narcotics are  sold, both the wholesale and the street value
... [and] ...  the Metropolitan Police Department and Drug Enforcement
 Administration procedures for the safeguarding of narcotics 
evidence." Brown went on to testify about the procedures  the Police
Department used to store narcotic substances and  to give his opinion,
in light of Williams' large collection of  small "baggies" of heroin,
that "[h]eroin users don't buy this  amount of dope broken down and
packaged like this for their  own personal use. It just does not


After Williams' conviction, his attorney learned that Brown  was not a
pharmacist and had no degree in pharmacology,  facts unknown to the
prosecution during the trial. Williams  then moved for a new trial
pursuant to Federal Rule of  Criminal Procedure 33, which the district
court denied.


What is the standard for ordering a new trial when the  newly
discovered evidence is that perjury occurred? Rule 33  says only: "the
court may grant a new trial ... if the  interests of justice so
require." An ancient opinion from  another circuit lays down this
test: a defendant is entitled to  a new trial if, without the perjured
testimony, "the jury might  have reached a different conclusion."
Larrison v. United  States, 24 F.2d 82, 87 (7th Cir. 1928). Notice
that the  Larrison formulation focuses on the importance of the per-
jured testimony to the prosecution's case. It does not ask  whether
the jury would have reached a different conclusion  had the perjury
been revealed at trial, although the Seventh  Circuit has now modified
the test to take this into account.  See United States v. Mazzanti,
925 F.2d 1026, 1030 & n.6 (7th  Cir. 1991). Notice too that Larrison
puts the test in terms of  what "might" have happened rather than what


Under our usual Rule 33 standard, a defendant is not  entitled to a
retrial on the basis of newly discovered evidence  unless he can show
that "a new trial would probably produce  an acquittal." United States
v. Thompson, 188 F.2d 652, 653  (D.C. Cir. 1951) (emphasis added).
This formulation, common  throughout the federal courts, has been used
for nearly a  century and a half. See 3 Charles Alan Wright, Federal 
Practice and Procedure s 557 (2d ed. 1982). We have  consistently
followed the Thompson standard in evaluating  motions for a new trial
under Rule 33. See United States v.  Gloster, 185 F.3d 910, 914 (D.C.
Cir. 1999). The difference  between Larrison and Thompson is not just
in the use of  "might" versus "probably." Thompson looks ahead and
eval- uates the outcome of a new trial; Larrison looks back and 
evaluates the impact of the perjury on the jury in the original 


This circuit has never adopted Larrison. In the past we  have managed
to avoid choosing between it and the standard  of Thompson because the
defendant was not entitled to a new  trial under either formulation.
See United States v. Mangi- eri, 694 F.2d 1270, 1286 (D.C. Cir. 1982);
United States v.  Mackin, 561 F.2d 958, 961 (D.C. Cir. 1977). Today we
join  several other circuits in rejecting Larrison. See United 


States v. Sinclair, 109 F.3d 1527, 1532 (10th Cir. 1997);  United
States v. Provost, 969 F.2d 617, 622 (8th Cir. 1992);  United States
v. Krasny, 607 F.2d 840, 844-45 (9th Cir. 1979);  United States v.
Stofsky, 527 F.2d 237, 246 (2d Cir. 1975). The  First Circuit in
United States v. Huddleston, 194 F.3d 214,  219 (1st Cir. 1999), also
refused to follow Larrison partly on  the basis of United States v.
Agurs, 427 U.S. 97, 103 (1976).  The Supreme Court there directed
federal courts to overturn  convictions based on the government's
knowing use of per- jured testimony if there is "any reasonable
likelihood that the  false testimony could have affected the judgment
of the jury."  Id. From this the First Circuit reasoned: "If courts
must  scrutinize the knowing use of perjured testimony under this 
standard, there is no principled justification for treating the 
government more harshly (such as by interposition of the  Larrison
rule) when its use of perjured testimony is inadver- tent."
Huddleston, 194 F.3d at 220. We are not so sure.  The Agurs test,
which repeats prior Supreme Court law, see  Giglio v. United States,
405 U.S. 150, 154-55 (1972); Napue  v. Illinois, 360 U.S. 264, 269-70
(1959), is quite easily satis- fied. The phrases--"reasonable
likelihood," "could have af- fected"--"mandate a virtual automatic
reversal of a criminal  conviction." Stofsky, 527 F.2d at 243. It is
hard to see how  Larrison could have set down an even more liberal


This is not to say that the Larrison test is difficult to  satisfy. It
is not. The Second Circuit's Stofsky opinion put  the matter well:
"the test, if literally applied, should require  reversal in cases of
perjury with respect to even minor  matters, especially in light of
the standard jury instruction  that upon finding that a witness had
deliberately proffered  false testimony in part, the jury may
disregard his entire  testimony." 527 F.2d at 245-46. That is reason
enough to  reject Larrison.


If not Larrison, what should the standard be? One possi- bility is the
standard laid down in Thompson for other types  of newly discovered
evidence. This would mean that, when  perjury by a prosecution witness
is discovered after trial and  when the prosecution did not know of
the perjury until then, 


a defendant would be entitled to a new trial only if he can  establish
that he would probably be acquitted on retrial.  History provides a
reason for adhering to the Thompson  formulation. Rule 33's current
text was adopted in 1944.  The accompanying Advisory Committee note
stated that the  rule "substantially continues existing practice."
Fed. R. Crim.  P. 33, advisory committee's note. The widely-accepted
prac- tice in 1944, a practice derived from a mid-19th century state 
court decision, see 3 Wright, supra, s 557, at 315, 322,  required a
defendant seeking a new trial to demonstrate a  likelihood of success
in a future retrial. See Evans v. United  States, 122 F.2d 461, 468-69
(10th Cir. 1941); Wagner v.  United States, 118 F.2d 801, 802 (9th
Cir. 1941); Prisament  v. United States, 96 F.2d 865, 866 (5th Cir.
1938); Johnson v.  United States, 32 F.2d 127, 130 (8th Cir. 1929).
Larrison too  predated Rule 33, but it had not been adopted in any
other  circuit, and in fact had been cited only twice in the other 
courts of appeals, and then only for propositions having  nothing to
do with this case. See Dale v. United States, 66  F.2d 666, 667 (7th
Cir. 1933); Vause v. United States, 54 F.2d  517 (2d Cir. 1931).


Another reason for adhering to the Thompson standard is  that
newly-discovered evidence of perjury is not distinguish- able from
other newly-discovered evidence. One author dis- agrees, arguing that
perjury is different because it creates  "an error at trial" whereas
in the case of other types of newly  discovered evidence, "the
evidence at trial may have been  incomplete, but it was all true."
Note, I Cannot Tell A Lie:  The Standard for New Trial in False
Testimony Cases, 83  Mich. L. Rev. 1925, 1945 (1985). The difference
is illusory.  Newly discovered evidence may often tend to prove that
the  evidence before the jury was not "true." A third party may 
confess to the crime; it may turn out that the main govern- ment
witness has a string of felony convictions; proof positive  of the
defendant's alibi might surface. Any one of these  items of newly
discovered evidence, in various degrees,  throws doubt on the accuracy
of the trial evidence pointing to  the defendant's guilt. Yet the


33 motions in such cases, will evaluate the motions by using  the
Thompson test.


We recognize that the Second Circuit in Stofsky, while  refusing to
follow Larrison, may have devised a variation of  it. Rather than
asking whether the outcome of the trial  might have been different had
the jury known of the witness's  lie, the Second Circuit asks whether
the defendant probably  would have been acquitted. This differs from
our Thompson  standard because, like Larrison, it looks at the matter
retro- spectively. The retrospective-prospective difference may not 
matter in the mine run of cases. But we can imagine  situations in
which it would matter, situations in which Stof- sky would command a
new trial that in all probability would  not produce a difference
outcome. Because we can see no  good reason to treat newly-discovered
evidence of perjury  differently than other types of newly-discovered
evidence, we  reject Stofsy and adhere to our original formulation


If Williams were retried, the government would have at its  disposal
any number of experts who could testify that the  amount of heroin in
his possession was inconsistent with  personal use. Or the government
could decide not to call an  expert on this subject. Any rational
juror could infer from  the fact that Williams was carrying 725
individual "baggies"  of heroin that he was intending to sell them.
See United  States v. Askew, 88 F.3d 1065, 1070 (D.C. Cir. 1996) (jury
 could infer from the quantity of drugs possessed that a  defendant
intended to distribute them even without expert  witness). So too the
fact that drugs were segregated into  "baggies" supports an inference
of intent to distribute. See  United States v. Glenn, 64 F.3d 706,
711-12 (D.C. Cir. 1995)  (drugs segregated into 9 "baggies" supports
inference of  intent to distribute). In either event--a different
expert or  no expert--it is most unlikely that a jury would acquit 


Affirmed.