A-1-CA-36490 | Omnilex

Texte intégral

Court of Appeals of New Mexico
Filed 7/24/2019 10:57 AM

OCR transcription for search; preserved printed pages are authoritative

—

10
1
12
13
14
15
16

Court of Appeals of New Mexico
Filed 7/24/2019 10:57 AM

IN THE COURT OF APPEALS OF THE STATE OF NEW wad 2
ark Reynolds

STATE OF NEW MEXICO,
Plaintiff-Appellee,
Vv. No. A-1-CA-36490
JEFFREY MORRILL,
Defendant-Appellant.
APPEAL FROM THE DISTRICT COURT OF BERNALILLO COUNTY

Briana H. Zamora, District Judge

Hector H. Balderas, Attorney General
Santa Fe, NM
Lauren J. Wolongevicz, Assistant Attorney General

Albuquerque, NM
for Appellee
Bennett J. Baur, Chief Public Defender
Kimberly Chavez Cook, Assistant Appellate Defender
Santa Fe, NM
for Appellant
MEMORANDUM OPINION
BOGARDUS, Judge.
4} Defendant appeals his conviction, following a bench trial, of two counts of
sexual exploitation of children (distribution), contrary to NMSA 1978, Section 30-

6A-3(B) (2007, amended 2016).' This crime is commonly referred to as distribution
of child pornography. Defendant advances four arguments: (1) there was insufficient
evidence to support his convictions for distribution; (2) his convictions of multiple
counts of distribution violate double jeopardy; (3) the district court abused its
discretion in finding the Roundup Torrential Downpour (Roundup) software to be
reliable; and (4) the district court erred in admitting the evidence produced by
Roundup and Forensic Toolkit. Persuaded only by Defendant’s double jeopardy
argument, we remand to the district court to vacate one of the two counts of
distribution and otherwise affirm.

BACKGROUND

{2} Defendant’s charges stem from his use of BitTorrent, a peer-to-peer, file-
sharing network, to access child pornography. The evidence at trial established the
following. While there is nothing inherently illegal about peer-to-peer, file-sharing
networks, they are the most common way to download or distribute child
pornography. BitTorrent is one of the most popular peer-to-peer, file-sharing
networks. People can connect to the BitTorrent network through a number of

different software programs, including Torrent, which is what Defendant used.

'Defendant was also convicted of one count of sexual exploitation of children
(possession), contrary to Section 30-6A-3(A). Defendant does not challenge this
conviction in this appeal.

433 Unlike users of other peer-to-peer, file-sharing networks, BitTorrent users do
not search for files within p Torrent. Rather, BitTorrent users must search websites
that act as a directory in order to obtain torrent files, which can contain a single file
or a number of files. Once a BitTorrent user locates a torrent file they would like to
download, the user downloads an info hash file. The info hash file tells p Torrent
what the user is looking for and where it can be found. Once the info hash is
downloaded, Torrent connects to Internet Protocol (IP) addresses that have been
identified as having the requested torrent file until it is able to download all or part
of the file. The more people on the peer-to-peer, file-sharing network who have a
torrent file, the faster that file can be downloaded.

{4 One of the State’s witnesses testified that peer-to-peer, file-sharing networks,
including BitTorrent, operate on the same protocol—‘to download, you have to
share.” In furtherance of that protocol, uTorrent is configured to automatically share
anything that a user downloads as long as it remains in the shared folder or in any
other folder the user designates to be shared. There is no way to turn off the sharing
feature in Torrent. If a pTorrent user is not sharing, the network notices and can
put the user’s software in a temporary state that does not allow the user to download
from others.

{5} Special Agent Owen Pefia with the New Mexico Internet Crimes Against

Children Task Force was qualified as an expert in peer-to-peer and peer-to-peer

investigations and testified to the following. Agent Pefia conducts investigations on
the BitTorrent network using Roundup. Agent Pefia’s computer runs Roundup
twenty-four hours a day, seven days a week. Roundup looks for particular torrent
files that are identified based on their unique info hash.

£6} If an IP address believed to be in New Mexico is sharing one of those
particular torrent files, the Roundup software used by Agent Pefia attempts to make
a connection to that IP address. In order to make a connection, three things must be
true: (1) the sharing computer is on; (2) the file-sharing software is running on the
sharing computer; and (3) the user has to be sharing at least a part of the identified
torrent file. If Roundup is able to make a connection to the identified IP address, it
attempts to download the identified torrent file. Unlike Torrent and other peer-to-
peer, file-sharing software, Roundup only downloads from a single source—the
specific IP address it has identified. Roundup logs every action it takes.

{7} On September 11-12 and October 2, 2015, the Roundup software used by
Agent Pefia connected to an IP address and downloaded a total of eight distinct
torrent files, which contained a total of 10,867 individual files. Once the torrent files
were finished downloading, Agent Pefia reviewed them and went through all of their
file structures. Upon finding child pornography, Agent Pefia initiated further

investigation.

{8} Qualified as an expert in peer-to-peer investigations and peer-to-peer systems,
Detective Kyle Hartsock with the Bernalillo County Sheriff's Department testified
to the following. Detective Hartsock received a case packet from Agent Pefia that
included a paper report and a CD/DVD that contained the actual files downloaded
and the log reports that showed the dates, times, and how the software was operating.
Based on the information contained in the case packet, Detective Hartsock obtained
and executed a search warrant for an apartment located in Albuquerque, New
Mexico. Following the search of the apartment, a number of electronic items,
including Defendant’s computer and external hard drives, were seized. Detective
Hartsock took the seized items to the Regional Computer Forensics Laboratory
where they underwent forensic examination using Forensic Toolkit.

{9}. Defendant was present when the search warrant was executed on the
apartment.” Several minutes after the search began, Detective Hartsock interviewed
Defendant. The interview lasted approximately fifty minutes and was recorded on
camera. The video of the interview was played at trial.

{10} During the interview, Defendant made a number of admissions and denials.
When Detective Hartsock told him why they were conducting the search, Defendant

admitted to having the images they were looking for on one or more of his external

*Defendant’s roommate was also present during the search. When he was
interviewed, Defendant reported that, to his knowledge, his roommate was not
involved, which was corroborated by the investigation.

hard drives. Defendant followed this statement by saying that it was his
understanding that the images were not child pornography.

{113 Defendant reported that the text associated with the images indicated that they
were of an artistic nature and were not meant for pornographic purposes. Defendant
also stated that he looked up the relevant laws and the images did not seem to meet
the statutes. Defendant admitted to downloading pTorrent to his computer, using
uTorrent to download the files, and setting up p Torrent to save the downloaded files
directly to his external hard drives. However, Defendant denied being aware that the
images were available to upload.

{123 Defendant admitted to searching for “LS,” but reported that it was his
understanding that the LS series’ was not child pornography. Defendant admitted to
viewing pictures from the LS series for over twenty years. On his external hard drive
in a “/Torrents folder,” Defendant created a folder titled “L” where he put the
images. Defendant stated child pornography charges were brought in different
countries against those involved with the LS series, but stated that all charges were

dropped and did not result in trials or convictions. Based on this information and the

3Both Agent Pefia and Detective Hartsock were familiar with the LS series
through their experience in child pornography investigations. Detective Hartsock
stated that a series can be dozens or hundreds of depictions with either the same
victim or the same location.

text associated with the images, Defendant reported that he did not believe that the
images were illegal.

4133. At Defendant’s bench trial, Detective Hartsock demonstrated how to install
the same version of pTorrent that Defendant used onto a computer. The
demonstration showed that an end user license agreement (EULA) must be agreed
to in order to install pTorrent. While a user does not have to scroll through the EULA
before agreeing, the agreement did contain a section titled “automatic uploading”
that explained that using uTorrent to download files would allow other users to
access those downloaded files. When the installation was complete, pTorrent opened
and then the window closed; however, the program remained running in the task bar.
Detective Hartsock explained that » Torrent was designed to be running all the time.
{14}. Detective Hartsock then demonstrated how to obtain a torrent file by going to
one of the online directories and finding an e-book. When he clicked on the
download link for the e-book, uTorrent opened to confirm what was to be done with
the file, including where to download the file. As the e-book downloaded, it was
shown in a downloading folder in Torrent that indicated information like the file
name, file size, download speed, upload speed, and estimated time the download
would finish. Once the e-book was completely downloaded, the file was no longer
shown in the downloading folder and instead showed in pTorrent’s seeding file,

indicating it was available to be shared.

{15} Defendant moved for a directed verdict at the close of the State’s evidence.
The district court granted the motion in part, but allowed two counts of distribution
of child pornography and one count of possession of child pornography to go
forward. Defendant did not put on any evidence, but renewed his directed verdict
motion after the defense rested. The district court denied the motion and ultimately
convicted Defendant of all remaining counts.

{16} Because this is a memorandum opinion and the parties are familiar with the
facts and procedural history of this case, we reserve discussion of additional
pertinent facts for our analysis.

DISCUSSION

I. Sufficiency of the Evidence

{17}. Relying on the analysis found in State v. Granillo, 2016-NMCA-094, 384
P.3d 1121, Defendant asks that we determine that general criminal intent is
insufficient to convict for distribution of child pornography under Section 30-6A-
3(B). Defendant argues, in the alternative, that the intent evidence was insufficient
even if we conclude that the mens rea is general intent.

{18}. We recently rejected a nearly identical argument for a heightened mens rea
under Section 30-6A-3(B) in State v. Franco, 2019-NMCA-___, 9.13, __ P.3d
___ (No. A-1-CA-35470, June 13, 2019). Like Defendant, the Franco defendant

also relied on Granillo in urging this Court to determine that general criminal intent

was insufficient under Section 30-6A-3(B). Franco, 2019-NMCA-__, § 13. We
reasoned that Section 30-6A-3(B) could not be said to “lean[] away from the
common law approach” like the statute involved in Granillo, NMSA 1978, § 30-6-
1(D)(1) (2009) (criminalizing intentional child abuse by endangerment), because
they did not have the same tiered structure. Franco, 2019-NMCA-___, 4 16.
Therefore, under the common law, we concluded Section 30-6A-3(B) only requires
general criminal intent because it “only describes a particular act and does not
include an intent to do a further act or achieve a further consequence.” Franco, 2019-
NMCA-__,4 16.

{19} For that reason, we analyze whether there was sufficient evidence presented
at trial that Defendant acted with general criminal intent. “The test for sufficiency of
the evidence is whether substantial evidence of either a direct or circumstantial
nature exists to support a verdict of guilty beyond a reasonable doubt with respect to
every element essential to a conviction.” State v. Montoya, 2015-NMSC-010, 52,
345 P.3d 1056 (internal quotation marks and citation omitted). “[S]ubstantial
evidence means such relevant evidence as a reasonable mind might accept as
adequate to support a conclusion[.]” State v. Salgado, 1999-NMSC-008, § 25, 126
N.M. 691, 974 P.2d 661 (internal quotation marks and citation omitted). “In
reviewing the sufficiency of the evidence, we must view the evidence in the light

most favorable to the guilty verdict, indulging all reasonable inferences and

18
19

resolving all conflicts in the evidence in favor of the verdict.” State v. Cunningham,
2000-NMSC-009, § 26, 128 N.M. 711, 998 P.2d 176.

20} “The element of general criminal intent is satisfied if the [s]tate can
demonstrate beyond a reasonable doubt that the accused purposely performed the
act in question.” State v. Gonzalez, 2005-NMCA-031, § 23, 137 N.M. 107, 107 P.3d
547 (alterations, internal quotation marks, and citation omitted); see UJI 14-141
NMRA. Here, the evidence established the following. Defendant admitted to
downloading pTorrent, which he used to access BitTorrent, a peer-to-peer, file-
sharing network. The BitTorrent network and Torrent required Defendant to share
to be able to continue to download files from others. Defendant chose to use
uTorrent, which does not allow users to turn off the sharing feature. Defendant used
uTorrent to download images from the LS series, a known child pornography series.
Defendant kept the child pornography he downloaded in a shared folder that allowed
others on the BitTorrent network to access it. The pTorrent software moved files
that were fully downloaded into a folder titled “seeding,” which indicated that the
files were being shared. Based on this evidence, we conclude there was substantial
evidence that Defendant acted with the requisite intent.

II. Defendant’s Multiple Convictions for Distribution Violate Double
Jeopardy

21; The district court convicted Defendant of two counts of distribution of child

pornography contrary to Section 30-6A-3(B). Defendant argues, and the State

1]
12

concedes, that State v. Sena, 2016-NMCA-062, 376 P.3d 887, requires us to vacate
all but one count. While we are not bound by the State’s concession, State v. Tapia,
2015-NMCA-048, § 31, 347 P.3d 738, we accept that concession because it is based
on binding precedent. See Sena, 2016-NMCA-062, §§ 13-19 (concluding a
defendant who was convicted under similar circumstances could only be convicted
of one count of distribution of child pornography under Section 30-6A-3(B)).
Accordingly, under the facts and circumstances of this case, we hold that
Defendant’s two convictions for distribution of child pornography violate double
jeopardy and must be reduced to a single conviction. See Sena, 2016-NMCA-062,
q 19.

III. The District Court Did Not Abuse Its Discretion in Finding Roundup
Reliable

{22} Defendant argues that the district court abused its discretion in finding the
technical conclusions reached by Roundup to be reliable. Defendant makes four
contentions: (1) the development and testing process of Roundup did not meet
industry standards; (2) any peer review Roundup was subject to was inadequate; (3)
Roundup’s error rate is unknown; and (4) Roundup does not incorporate any
accuracy-ensuring measures.

23} “We review evidentiary decisions by the district court for an abuse of
discretion.” State v. Nichols, 2006-NMCA-017, 4 20, 139 N.M. 72, 128 P.3d 500.

“An abuse of discretion occurs when the ruling is clearly against the logic and effect

of the facts and circumstances of the case.” State v. Flores, 2010-NMSC-002, § 25,
147 N.M. 542, 226 P.3d 641 (internal quotation marks and citation omitted). “We
cannot say the [district] court abused its discretion by its ruling unless we can
characterize it as clearly untenable or not justified by reason.” /d. (internal quotation
marks and citation omitted).

(24 Three requirements must be met before evidence is admissible under Rule 11-
702 NMRA, which governs the admissibility of expert testimony: “(1) [the] experts
must be qualified; (2) their testimony must assist the trier of fact; and (3) their
testimony must be limited to the area of scientific, technical, or other specialized
knowledge in which they are qualified.” State v. Torres, 1999-NMSC-010, § 23, 127
N.M. 20, 976 P.2d 20. Here, Defendant’s challenge is to the third requirement, which
involves a determination of the reliability of scientific evidence. See State v.
Alberico, 1993-NMSC-047, J 47, 116 N.M. 156, 861 P.2d 192 (“When scientific
evidence is employed as a means of obtaining or analyzing data, the [district] court
must determine whether the scientific technique is based upon well-recognized
scientific principle and whether it is capable of supporting opinions based upon
reasonable probability rather than conjecture.”).

725} Because “evidentiary reliability is the hallmark for the admissibility of
scientific knowledge[,]” Torres, 1999-NMSC-010, § 26, “the party offering expert

testimony based on scientific knowledge must establish that such knowledge is not

—

CoO WANN HS

only relevant, but reliable.” State v. Yepez, 2018-NMCA-062, § 21, 428 P.3d 301,
cert. granted, 2018-NMCERT-__ (No. S-1-SC-37217, Sept. 28, 2018). To
determine reliability, the following factors should be considered:
(1) whether a theory or technique can be (and has been) tested; (2)
whether the theory or technique has been subjected to peer review and
publication; (3) the known [or] potential rate of error in using a
particular scientific technique and the existence and maintenance of
standards controlling the technique's operation; and (4) whether the
theory or technique has been generally accepted in the particular
scientific field.
Id. § 22 (internal quotation marks and citation omitted). “In addition to these four
Daubert factors, New Mexico courts rely upon a fifth factor: whether the scientific
technique is capable of supporting opinions based upon probability rather than
conjecture.” /d. (omission, internal quotation marks, and citation omitted); see
Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993); Alberico, 1993-NMSC-
047, 9 47.
{26} The district court held a Daubert-Alberico hearing over the course of two
days. Three witnesses testified in regard to the reliability of Roundup—Detective
Robert Erdely and Agent Pefia for the State and Thomas Blog for Defendant. At the
conclusion of the hearing, the district court found that the State had met its burden
on reliability and that Defendant’s arguments went to the weight of the evidence.

After careful review of the testimony of the witnesses, we conclude that the district

court did not abuse its discretion when it determined that the evidence generated by

Roundup was sufficiently reliable to be admitted at trial. We summarize the
witnesses’ testimony as follows.

27; Mr. Blog testified regarding his concerns with Roundup’s reliability in
general, including that the software had not been appropriately tested, had not been
subjected to formal peer review, and that its error rate had not been specifically
established. However, Detective Erderly testified that the software had been
independently tested one time to ensure that the program does single source
downloads, it was not sharing, and it was keeping accurate logs. The independent
test was done by an unidentified person hired by the FBI and involved a review of
Roundup’s source code, black-box testing, and a determination of the rate of failure.
The independent test used Wireshark, a widely accepted form of monitoring the flow
of information over an internet cable, to evaluate Roundup’s results. Detective
Erdely reported that the independent tester concluded that Roundup performed as
described. Detective Erdely also testified that Roundup’s results are validated at the
end of every training that he teaches on the program and through every investigation
that is undertaken with it. Mr. Blog acknowledged that real-world testing is a form
of black-box testing.

(283 Detective Erdely conceded that Roundup had not been subjected to academic
peer review outside of the University of Massachusetts Amherst, where the program

was developed. He noted, however, that every investigator that uses Roundup does

black-box testing and validation, which is a form of peer review. Mr. Blog testified
that Roundup’s change logs were not public, which would have shown what bugs
have been fixed, what bugs remain, and what operating systems and environments
Roundup had been tested on. Detective Erdely explained that the release of software
source code is not done for law enforcement tools, which also explains why change
logs are not made available.

29} Addressing concerns regarding Roundup’s error rate, Detective Erdely stated
that false positives are prevented because the program uses SHA-1 hashing, which
compares two groups of information to see if they are exactly the same. Detective
Erdely has never seen Roundup return a false positive during any of his in-class
validations, nor has he received any reports of false positives from the 1,000 plus
law enforcement personnel who use the program worldwide. Agent Pefia testified
that he had been using Roundup for over two years and that it had a zero percent
error rate in his experience. Agent Pefia also testified that he completed an in-class
validation at the end of his training on Roundup, which demonstrated a zero percent
rate of error.

{30} Detective Erdely testified that the log created by Roundup, which documents
the time actions are taken, what torrent is being investigated, the suspect computer’s
IP address, the suspect’s computer’s port, the version of the software used by the

suspect computer, and the investigator’s IP address, is equivalent to Wireshark

18
19

monitoring. Detective Erdely opined that Roundup’s logs are even better than
Wireshark in some aspects because the suspect computer is the computer identifying
both the investigator’s IP address and its own IP address.

31). Based on the foregoing, we cannot conclude that the district court’s
determination that Roundup was reliable to be clearly untenable or not justified by
reason. See Flores, 2010-NMSC-002, § 25. While it does not appear that Roundup
has not been subjected to peer review, and Defendant presented expert testimony
that pointed out potential problems with the reliability of Roundup, the State
provided sufficient evidence from which the district court could reasonably conclude
that Roundup was sufficiently reliable based on an analysis of all of the Daubert-
Alberico factors. We agree with the district court that Defendant’s arguments go to
the weight of the evidence. See Acosta v. Shell W. Expl. & Prod., Inc., 2016-NMSC-
012, § 28, 370 P.3d 761 (“Vigorous cross-examination, presentation of contrary
evidence, and careful instruction on the burden of proof are the traditional and
appropriate means of attacking shaky but admissible evidence.” (internal quotation
marks and citation omitted)). For these reasons, we cannot say the district court
abused its discretion.

IV. The District Court Did Not Err in Admitting Evidence Produced by
Roundup and Forensic Toolkit

(32) . Over Defendant’s objection, the district court admitted the evidence produced

by Roundup after concluding the evidence was not hearsay because it was computer

generated. Similarly, the district court admitted the evidence produced by Forensic
Toolkit after concluding that it was admissible under several theories: that it was not
hearsay either because it was computer generated or not offered for the truth of the
matter asserted or, alternatively, if it was hearsay, there was adequate foundation for
admission under the business record exception. The district court did note that the
bookmarks and comments from the officers contained on the evidence produced by
Forensic Toolkit were hearsay and would not be considered. “[T]he threshold
question of whether the [district] court applied the correct evidentiary rule or
standard is subject to de novo review on appeal.” Torres, 1999-NMSC-010, § 28.
{33} Hearsay is defined as “a statement that (1) the declarant does not make while
testifying at the current trial or hearing, and (2) a party offers in evidence to prove
the truth of the matter asserted in the statement.” Rule 11-801(C) NMRA. A
“(d]jeclarant” is defined as “the person who made the statement.” Rule 11-801(B)
(emphasis added). A “[s]tatement” is defined as “a person’s oral assertion, written
assertion, or nonverbal conduct, if the person intended it as an assertion.” Rule 11-
801(A) (emphasis added). Whether the computer-generated evidence produced by
Roundup and Forensic Toolkit is hearsay depends on whether the evidence
constitutes statements under our Rules of Evidence.

(34; Defendant asks this Court to “find that the attestations made by a computer

program constitute ‘statements,’ whether attributable to an artificial intelligence

software or the software developer who implicitly offers the program’s conclusions
as their own.” (Emphasis omitted.) Based on that contention, Defendant further
argues that the automated conclusions from Roundup and Forensic Toolkit constitute
inadmissible hearsay statements that are not admissible under the business record
exception.’ In so arguing, Defendant acknowledges that such a holding would
diverge from the plain language of our hearsay rule’s relevant definitions that
reference statements of a “person.” Defendant does not cite any authority that
supports his argument and instead cites out-of-state authority that discussed and
rejected a similar argument. “We are entitled to assume, when arguments are
unsupported by cited authority, that supporting authorities do not exist.” State v.
Ibarra, 1993-NMCA-040, 4 13, 116 N.M. 486, 864 P.2d 302.

{35}. Based on the following, we conclude the district court correctly determined
that the computer generated evidence produced by Roundup and Forensic Toolkit
was not hearsay. Agent Pefia testified that his computer runs Roundup twenty-four

hours a day, seven days a week and automatically attempts to make connections with

‘Defendant also argues that we should adopt a scientific reliability test for
“out-of-court statements” made by Forensic Toolkit if we determine they are not
governed by our hearsay rules. Defendant does not develop this argument. Rather,
in conclusory and circular fashion, Defendant contends that Forensic Toolkit is
unreliable because the State did not qualify an expert to discuss it. We decline to
address this undeveloped argument. See State v. Guerra, 2012-NMSC-014, § 21,
278 P.3d 1031 (stating that appellate courts are under no obligation to review unclear
or undeveloped arguments).

and downloads from IP addresses that are suspected to be sharing child pornography.
As it does so, Roundup logs every action it takes. Detective Hartsock testified that
Forensic Toolkit organizes information stored on seized electronic devices into
various categories including graphics, videos, word documents, and internet history.
Because the software programs make the relevant assertions, without any
intervention or modification by a person using the software, we conclude that the
assertions are not statements by a person governed by our hearsay rules.°

36} Our conclusion is supported by the purpose behind our hearsay rule. “The
hearsay rule excludes from admissible evidence statements that are inherently
untrustworthy because of the risk of misperception, failed memory, insincerity,
ambiguity, and the like.” State v. Mendez, 2010-NMSC-044, 9 19, 148 N.M. 761,
242 P.3d 328. Statements produced by software are not subject to the same types of

risks.° Because the evidence produced by Roundup or Forensic Toolkit does not

The only exceptions are the bookmarks and comments in Forensic Toolkit,
which the district court did not consider.

°Evidence produced by software presents other potential risks. One such risk
is a malfunction that could result in misleading data, which can be addressed through
our authentication rule. See Rule 11-901(A) NMRA (“To satisfy the requirement of
authenticating or identifying an item of evidence, the proponent must produce
evidence sufficient to support a finding that the item is what the proponent claims it
is.”). Another potential risk is the production of unreliable data, which can be
addressed under Rule 11-702.

[o,¢)

1]
12

13
14

constitute hearsay, we conclude that the district court did not err in admitting the
same.

CONCLUSION

37 We remand to the district court with instructions to vacate one of Defendant’s

two convictions for distribution of child pornography. We otherwise affirm

Defendant’s convictions.
3833} ITISSO ORDERED.
his tro Foyt bad
KRISTINA BOGARDUS, Judge
WE CONCUR:

M Neng Spee

M. MONICA ZAMORA, hief Judge

ZACHARKA. VES: Judge

July 24, 2019- State of New Mexico v. Jeffrey Morrill, No. A-1-CA-36490

Texte intégral

OCR transcription for search; preserved printed pages are authoritative

Poursuivez vos recherches dans ChatGPT ou Claude