In the
United States Court of Appeals
For the Seventh Circuit
____________________
No. 20-2451
U NITED STATES OF A MERICA ,
Plaintiff-Appellee,
v.
WILLIAM A. J ULIUS ,
Defendant-Appellant.
____________________
Appeal from the United States District Court for the
Eastern District of Wisconsin.
No. 19-cr-116 — William C. Griesbach, Judge.
____________________
A RGUED SEPTEMBER 9, 2021 — D ECIDED SEPTEMBER 24, 2021
____________________
Before KANNE , HAMILTON, and ST . EVE , Circuit Judges.
ST . EVE , Circuit Judge. A jury found that William Julius set
fire to the building where his ex-girlfriend was living after she
spurned his attempts to rekindle their relationship. On appeal
Julius argues that the district court erred in allowing lay wit-
nesses to offer expert testimony about the process of extract-
ing data from his cellphone and in cutting off his cross-exam-
ination of one of those witnesses. We find no reversible errors
and affirm.
-- 1 of 10 --
2 No. 20-2451
I. Background
A federal grand jury charged Julius with two counts of ar-
son for setting fire to a building where his ex-girlfriend was
living, twice in the same night. Julius went to trial on both
counts.
The evidence at trial showed that Julius wanted to salvage
his relationship with his ex-girlfriend, Dawn Noack, but No-
ack was uninterested. The government’s theory was that Jul-
ius set the fires in retaliation for Noack’s rejection of his en-
treaties. At the time, Noack was living with her friend, Maro
Saldana, in Saldana’s apartment. Shortly before the fires, Jul-
ius was hanging around Saldana’s apartment on his bike. Sal-
dana approached Julius, told him that Noack did not want to
see him, and asked him to leave. On another occasion shortly
before the fires, Julius threw rocks at the apartment window.
Saldana again asked him to leave.
To prove Julius’s motive for setting the fires, the govern-
ment introduced text messages between Julius and Noack in
the days leading up to the fires. The text messages, which law
enforcement extracted from Julius’s cellphone, showed No-
ack was upset that Julius would not leave her alone. Julius
messaged Noack about his “heartache” and physical affection
for her, but in response Noack told him to “quit” and “knock
it the fuck off” because “we are over” and “I don’t want you
with me!” Julius also expressed frustration that Noack was
living with Saldana, saying “I don’t want you there. They
don’t like me”; “They better get rid of you”; “It’s easier if they
throw you out the door”; and “You’re out the door.” On the
afternoon before the fires, Julius mentioned in texts to some-
one else that he was “sobering up.”
-- 2 of 10 --
No. 20-2451 3
On the night of the fires, Julius texted Noack repeatedly
between 9:45 p.m. and 10:15 p.m. asking if she was “coming
outside.” At 10:30 p.m., he called her five times in a row, but
she did not answer. After midnight, Saldana’s cousin, who
lived in the same building, woke up to the smell of smoke. He
searched for the source and found burning coals on the inside
of the building’s front door frame. Saldana’s cousin alerted
Saldana, who called the police. At 12:57 a.m., the police and
fire departments were dispatched to the fire. At 1:03 a.m., Jul-
ius called Noack. The call lasted just under a minute. When
the police and fire departments arrived at the building, the
fire was out but the door frame was still smoking. The captain
of the fire department testified that the burned area smelled
of gasoline. Testing confirmed the presence of gasoline.
Around 3:30 a.m., the fire captain noticed a larger fire on
the other side of the building. This fire, too, was contained,
but it caused significant damage to the first floor of the build-
ing. The captain concluded that someone started the fires by
putting an open flame, like a lighter, to combustibles and gas-
oline. While he did not smell gasoline near the second fire, he
concluded that gasoline “may or may not have been used” be-
cause the fire could have consumed the gasoline.
The police suspected Julius of setting the fires. Shortly af-
ter the second fire, an officer spotted a man on a bicycle sev-
eral blocks from the fires. Upon approaching, the officer
found Julius hiding under a car with the bicycle close at hand.
The officer patted Julius down and found a lighter in his front
pocket. Julius did not have any cigarettes or cigars on him.
Julius was “clearly intoxicated” but could walk and carry on
a coherent conversation. Testing revealed gasoline on Julius’s
shoes and socks.
-- 3 of 10 --
4 No. 20-2451
The government called two witnesses—a computer foren-
sic examiner with the state police and an agent with the
United States Bureau of Alcohol, Tobacco, Firearms, and Ex-
plosives (ATF)—to testify to the process of extracting text
messages from Julius’s phone. This testimony laid the foun-
dation for the text messages from Julius’s phone. The govern-
ment did not seek to qualify these witnesses as experts, and
Julius did not object on that basis. For reasons that are unclear,
the district court ultimately instructed the jury that both wit-
nesses had provided opinion, but not expert opinion, testi-
mony about “telephone extraction data.”
Beyond testifying to the extraction process, the ATF agent
testified about certain location data obtained from the extrac-
tion. On direct examination, she testified that she tried to de-
termine the location of Julius’s phone from the extraction
data, but that she was unable to “substantiate” the location
data in the extraction report and thus could not reach any con-
clusions as to the phone’s location. On cross-examination, de-
fense counsel followed up on the location data. He asked
about a specific data point from 1:09 a.m. on the night of the
fire. The agent testified that this data point did not provide
any information about the location of Julius’s cellphone that
night. She confirmed, though, that the data point corre-
sponded to a location at “North Clay and 141.” Before defense
counsel could ask further questions, the government objected.
The district court sustained the objection at a sidebar off the
record. On redirect, the agent reiterated that there was “no re-
liable location data” from the cellphone extraction. Later, and
back on the record, the court explained the basis for its ruling
limiting Julius’s cross-examination: The testimony had “no
foundation,” meaning it “wasn’t reliable so it was not admis-
sible as expert testimony or as evidence.”
-- 4 of 10 --
No. 20-2451 5
In addition to this evidence, the government introduced
several post-indictment letters between Julius and his proba-
tion officer.* In the letters Julius admitted that he had texted
Noack on the night of the fires to ask if she was coming out-
side. He also provided contradictory accounts of what tran-
spired that night, initially saying that Noack started the first
fire and later saying Saldana’s son started the fires. Julius
elected not to testify on his own behalf. The jury convicted
him on both counts. Julius now appeals.
II. Discussion
Julius maintains that the district court committed two ev-
identiary errors during trial. We conclude, however, that nei-
ther of the asserted errors affected the outcome of the trial. We
thus affirm Julius’s conviction.
A. Expert Testimony
Julius begins by challenging the district court’s failure to
qualify the forensic examiner and ATF agent as expert wit-
nesses before allowing them to testify to the process of ex-
tracting data from his cellphone. See Fed. R. Evid. 702; Daubert
v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993). He relies
chiefly on United States v. Wehrle, 985 F.3d 549 (7th Cir. 2021),
which held that a police officer’s testimony about her forensic
examination of external storage devices was expert testimony
because it involved “technical concepts beyond ordinary
knowledge.” Id. at 554. The devices at issue in Wehrle were not
cellphones, id. at 553 n.2, but Julius contends that Wehrle’s
* Julius was on federal supervised release in another case at the time of the
fires. The jury did not know this; they knew the probation officer only as
a person with whom Julius had a “professional relationship.”
-- 5 of 10 --
6 No. 20-2451
rationale applies with equal force to testimony about extract-
ing data from a cellphone. Other circuits have held that testi-
mony about extracting data from a cellphone is not expert tes-
timony. United States v. Montijo-Maysonet, 974 F.3d 34, 48 (1st
Cir. 2020); United States v. Chavez-Lopez, 767 F. App’x 431, 433–
34 (4th Cir. 2019); United States v. McLeod, 755 F. App’x 670,
673 (9th Cir. 2019); United States v. Marsh, 568 F. App’x 15, 17
(2d Cir. 2014).
We need not resolve this issue. Julius concedes that he did
not object to the district court’s failure to qualify the forensic
examiner and ATF agent as expert witnesses after the govern-
ment failed to tender them as experts, so we review for plain
error. See Fed. R. Crim. P. 52(b). To prevail on plain-error re-
view, Julius must show: (1) an error; (2) that is plain; (3) that
affected his substantial rights, meaning there is a reasonable
probability that the error changed the outcome of the trial;
and (4) that “had a serious effect on ‘the fairness, integrity or
public reputation of judicial proceedings.’” Greer v. United
States, 141 S. Ct. 2090, 2096–97 (2021) (quoting Rosales-Mireles
v. United States, 138 S. Ct. 1897, 1905 (2018)). Julius has the bur-
den of establishing all four elements, yet he concedes he can-
not satisfy the third element. More specifically, he concedes
he cannot show that the forensic examiner and ATF agent
were unqualified to give the challenged testimony. See United
States v. Thomas, 970 F.3d 809, 815 (7th Cir. 2020) (finding no
plain error where witness’s testimony would have been “un-
objectionable” had the government offered him as an expert);
see also Wehrle, 985 F.3d at 554. Julius disagrees with our prec-
edent on plain error in this context, but he does not presently
ask us to revisit it. Because there is no plain error, we leave
Wehrle’s broader implications for another day.
-- 6 of 10 --
No. 20-2451 7
B. Location Data
Next, Julius maintains that the district court improperly
denied him the opportunity to cross-examine the ATF agent
about the location data obtained through the cellphone ex-
traction. As mentioned, the court sustained the government’s
objection to Julius’s cross-examination at a sidebar off the rec-
ord. Later, the court explained on the record that it sustained
the objection because the testimony had “no foundation,”
meaning it “wasn’t reliable so it was not admissible as expert
testimony or as evidence.”
This reasoning is hard to follow. The court did not qualify
the ATF agent as an expert, so it had no basis for considering
the reliability of her testimony under the Daubert framework.
See Daubert, 509 U.S. at 590. And while foundation is an im-
portant evidentiary concern, it is unclear from the record why
the court concluded that the testimony lacked foundation. So,
the court’s post hoc summary of its off-the-record ruling does
not necessarily explain its reason for cutting off Julius’s cross-
examination.
The court’s decision to rule off the record also makes it
hard for us to know whether Julius preserved his request for
further cross-examination. It does not appear from the record
that Julius made an offer of proof, see Fed. R. Evid. 103(a)(2),
but we cannot be sure without knowing what transpired at
the sidebar. As we explained more than 30 years ago, the
Court Reporter’s Act, 28 U.S.C. § 753(b), “requires court re-
porters to record verbatim ‘all proceedings in criminal cases
had in open court,’” including sidebars. United States v. Nolan,
910 F.2d 1553, 1559 (7th Cir. 1990) (quoting § 753(b)). Without
a contemporaneous record of a sidebar discussion, it can be
difficult to discern the parties’ arguments or the basis for the
-- 7 of 10 --
8 No. 20-2451
court’s ruling. Id. at 1559–60. Once again, we remind district
courts to make evidentiary rulings on the record. See id. at
1560 (“The duty to comply with § 753(b) lies with the court,
not the parties.”).
The government says little in defense of the court’s ruling,
and instead maintains that the court’s error (if any) was harm-
less. See Fed. R. Crim. P. 52(a). “The test for harmless error is
whether, in the mind of the average juror, the prosecution’s
case would have been significantly less persuasive had the
improper evidence been excluded.” United States v. Brown, 973
F.3d 667, 707 (7th Cir. 2020) (quoting United States v. Emerson,
501 F.3d 804, 813 (7th Cir. 2007)). The government bears the
burden of demonstrating that an error was harmless. See
United States v. Jett, 982 F.3d 1072, 1078 (7th Cir. 2020).
In the government’s view, any error here was harmless be-
cause, at best, the location data from the cellphone extraction
would have shown that Julius was one mile away from the
fires at 1:09 a.m.—12 minutes after the police and fire depart-
ments were dispatched to the first fire. The government asks
us to take judicial notice of Google Maps data indicating that
it would take seven minutes to bike the one-mile distance be-
tween the fires and the intersection corresponding to the loca-
tion data. See Fed. R. Evid. 201(b).
We and other courts have taken judicial notice of distance
estimates from Google Maps. Cloe v. City of Indianapolis, 712
F.3d 1171, 1177 n.3 (7th Cir. 2013), overruled on other grounds by
Ortiz v. Werner Enters., Inc., 834 F.3d 760 (7th Cir. 2016); see also
United States v. Perea-Rey, 680 F.3d 1179, 1182 n.1 (9th Cir.
2012); Citizens for Peace in Space v. City of Colorado Springs, 477
F.3d 1212, 1219 n.2 (10th Cir. 2007). But travel-time estimates
are a different matter. Any number of factors could impact a
-- 8 of 10 --
No. 20-2451 9
cyclist’s travel time, including the cyclist’s level of intoxica-
tion (recall that Julius was drunk), the type and quality of the
bicycle, and the cyclist’s proficiency at riding a bike. The gov-
ernment does not grapple with these variables, so it has not
proven that Google Maps is a source “whose accuracy cannot
reasonably be questioned” for bike-time estimates. Fed. R.
Evid. 201(b)(2); see also Jackson v. Allstate Ins. Co., 785 F.3d 1193,
1204 (8th Cir. 2015) (declining to take judicial notice of Google
Maps drive-time estimate for similar reasons).
Even so, we are satisfied that any error the district court
made in limiting Julius’s cross-examination was harmless.
The government introduced powerful evidence of Julius’s
guilt. Julius was in a tumultuous relationship with Noack, an
occupant of the burned buildings. He had been hanging
around the building and throwing rocks at it in the days lead-
ing up to the fires. In text messages to Noack, Julius expressed
frustration that she was staying in the building, even making
vague threats such as “They better get rid of you”; “It’s easier
if they throw you out the door”; and “You’re out the door.”
On the night of the fires, Julius repeatedly texted and called
Noack, asking if she was coming outside. Julius called Noack
minutes after the first fire. Shortly after the second fire, a po-
lice officer found Julius hiding under a nearby car with gaso-
line on his shoes and socks and a lighter in his pocket. The fire
captain would later testify that both fires were caused by
someone applying an open flame, like a lighter, to combusti-
bles and gasoline. Meanwhile, in letters to his probation of-
ficer Julius provided contradictory accounts of who started
the fires.
Given this evidence, we are convinced that further cross-
examination about the location data from Julius’s phone
-- 9 of 10 --
10 No. 20-2451
would not have affected the jury’s verdict. Julius complains
that the court’s ruling prevented him from arguing to the jury
that he was a mile away from the site of the fires 12 minutes
after the first fire. With or without a Google Maps estimate,
though, a jury could easily infer that Julius could travel one
mile by bike in 12 minutes. Moreover, it seems unsurprising
that an arsonist might want to distance himself from a fire im-
mediately after setting it. On these facts, the government’s
case would not have been significantly less persuasive if the
court had allowed Julius to further cross-examine the ATF
agent about the location data. See Brown, 973 F.3d at 707.
A FFIRMED
-- 10 of 10 --