Jeffrey Masson’s odyssey through the federal courts continues. On remand from the Supreme Court, — U.S. -, 111 S.Ct. 2419, 2437, 115 L.Ed.2d 447 (1991), we consider what responsibility, if any, publishers have to purge defamatory material from stories submitted to them by free-lance writers.
Background
Masson was the subject of a not altogether flattering article written by Janet Malcolm, published by The New Yorker Magazine and republished in book form by Alfred A. Knopf. Masson alleges that the article libels him by attributing to him statements he did not make. He concedes he’s a public figure for purposes of this litigation, and so must meet the actual malice standard of New York Times Co. v. Sullivan, 376 U.S. 254, 279-80, 84 S.Ct. 710, 725-26, 11 L.Ed.2d 686 (1964), and Gertz v. Robert Welch, Inc., 418 U.S. 323, 342, 94 S.Ct. 2997, 3008, 41 L.Ed.2d 789 (1974). To avoid summary judgment under this standard, Masson was required to show that a reasonable jury could find by clear and convincing evidence that defendants published the statements with knowledge of their falsity or reckless disregard as to whether or not they were true. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255-56, 106 S.Ct. 2505, 2513-14, 91 L.Ed.2d 202 (1986); see also Sullivan, 376 U.S. at 279-80, 285-86, 84 S.Ct. at 725-26, 728-29.
*898The Supreme Court held that, when applied to altered quotations, falsity means “a material change in the meaning conveyed by the statement.” 111 S.Ct. at 2433. On this record, the Supreme Court held that five of the “quotations” of which Masson complains could meet this standard and thus are actionable. Id. at 2435-37.1 We must consider two questions: First, whether Masson’s case nevertheless fails against all defendants as a matter of state law under the so-called incremental harm doctrine; second, whether Masson has presented sufficient evidence against The New Yorker and Knopf to establish that each of them acted with actual malice (the Supreme Court having resolved this issue as to Malcolm).2
Discussion
I
As the Second Circuit has explained it, the incremental harm doctrine measures the harm “inflicted by the challenged statements beyond the harm imposed by the rest of the publication. If that harm is determined to be nominal or nonexistent, the statements are dismissed as not actionable.” Herbert v. Lando, 781 F.2d 298, 311 (2d Cir.), cert. denied, 476 U.S. 1182, 106 S.Ct. 2916, 91 L.Ed.2d 545 (1986). The defendants assert that the unchallenged or verifiably accurate statements attributed to Masson paint him in such a bad light that any damage caused by Malcolm’s allegedly creative quotations is not actionable as a matter of law. Cf. Richard N. Winfield, Altered Quotes and Incremental Harm, N.Y.St.Bar J. 16, 18 (Jan.1992) (encouraging courts “to attempt to comprehend the gestalt of a publication”).
In our earlier opinion we applied the incremental harm doctrine to one of the challenged quotations. See Masson v. New Yorker Magazine, Inc., 895 F.2d 1535, 1541 (9th Cir.1989) (citing Herbert and a district court opinion, Simmons Ford, Inc. v. Consumers Union, 516 F.Supp. 742 (S.D.N.Y.1981), on which Herbert relied). We did not, however, specify the source of the doctrine. Herbert noted that the theory was “novel,” and that it was first applied in Simmons. 781 F.2d at 310. Simmons, in turn, seemed to ground the doctrine in the First Amendment. 516 F.Supp. at 750-51. The Supreme Court, however, disowned this constitutional pedigree when it “rejected] any suggestion that the incremental harm doctrine is compelled as a matter of First Amendment protection for speech.” Masson, 111 S.Ct. at 2436. We must nevertheless consider whether it has an independent existence as a matter of state law.
The California courts have shown no interest in the incremental harm doctrine. In fact, we have found only a single reference to the doctrine in the reported California cases: In Weller v. American Broadcasting Cos., 232 Cal.App.3d 991, 1010, 283 Cal.Rptr. 644 (1991), defendant claimed the trial court erred in not instructing the jury on the incremental harm doctrine; because no incremental harm instruction had been offered, however, the court declined to address the issue. The parties have pointed us to no cases indicating that the California Supreme Court would be likely to adopt the doctrine, and we have found none.
In language we cannot improve upon, then-Judge Scalia expressed the fundamental flaw in the incremental harm doctrine:
[T]he theory must be rejected because it rests upon the assumption that one’s reputation is a monolith, which stands or falls in its entirety. The law, however, proceeds upon the optimistic premise *899that there is a little bit of good in all of us—or perhaps upon the pessimistic assumption that no matter how bad someone is, he can always be worse.... (“He was a liar and a thief, but for all. that he was a good family man.”)
Liberty Lobby, Inc. v. Anderson, 746 F.2d 1563, 1568 (D.C.Cir.1984), vacated on other grounds, 477 U.S. 242, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986); Cf. Wall St.J. A16 (Feb. 21, 1992) (psychiatrist describing Jeffrey Dahmer as “an organized, non-social, lust murderer,” but “not such a bad person”). Judge Scalia concluded that the incremental harm doctrine is simply a “bad idea.” Liberty Lobby, 746 F.2d at 1569.
Because it is not required by the First Amendment, because the Supreme Court has severely undermined the case authority that generated the doctrine in the first place, because the California courts have never adopted it and because we believe the California Supreme Court would agree with Judge Scalia that it is a “bad idea,” we conclude that the incremental harm doctrine is not an element of California libel law.
II
“[W]e must assume, except where otherwise evidenced by the transcripts of the tape recordings, that [Masson] is correct in denying that he made the statements attributed to him by Malcolm, and that Malcolm reported with knowledge or reckless disregard of the differences between what [Masson] said and what was quoted.” 111 S.Ct. at 2435. Furthermore, “[t]he record contains substantial additional evidence ... which, in a light most favorable to [Mas-son], would support a jury determination under a clear and convincing standard that Malcolm deliberately or recklessly altered the quotations.” Id. The law of the case then is that the evidence presented by Mas-son in opposition to summary judgment would support a jury verdict in his favor against Malcolm.3
The Supreme Court- has instructed us to determine whether “the District Court erred in granting summary judgment to The New Yorker Magazine, Inc., and Alfred A. Knopf, Inc. on the basis of their respective relations with Malcolm or the lack of any independent actual malice.” 111 S.Ct. at 2437. In making this determination we are guided by Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 109 S.Ct. 2678, 105 L.Ed.2d 562 (1989):
If a false and defamatory statement is published with knowledge of falsity or a reckless disregard for the truth, the public figure may prevail. A “reckless disregard” for the truth, however, requires more than a departure from reasonably prudent conduct. There must be sufficient evidence to permit the conclusion that the defendant in fact entertained serious doubts as to the truth of his 'publication. The standard is a subjective one—there must be sufficient evidence to permit the conclusion that the defendant actually had a high degree of awareness of probable falsity. As a result, failure to investigate before publishing, even when a reasonably prudent person would have done so, is not sufficient to establish reckless disregard. In a case such as this involving the reporting of a third party’s allegations, recklessness may be found where there are obvi-. ous reasons to doubt the veracity of the informant or the accuracy of his reports.
Id. at 688, 109 S.Ct. at 2696 (internal quotations, ellipses and citations omitted; emphasis added).
To overcome summary judgment (or sustain a jury verdict), therefore, a public official or public figure must establish by clear and convincing evidence that the publisher of an allegedly defamatory book or article “in fact entertained serious doubts *900as to the truth of his publication.” A plaintiff may do this in two ways. First, he can establish the publisher’s state of mind by evidence that the publisher “actually had a high degree of awareness of probable falsity.” Second, a plaintiff may show that the circumstances surrounding publication gave the publisher “obvious reasons to doubt the veracity of the [author] or the accuracy of [facts and quotations in her book or article],” and that the publisher failed to take reasonable steps to dispel those doubts.
As we interpret Harte-Hanks, these two standards provide different paths to the same conclusion. Where the jury has proof that a publisher “actually had a high degree of awareness of probable falsity,” that alone will establish that it “in fact entertained serious doubts as to the truth of [its] publication.” Where such direct proof is missing, the jury may nevertheless infer that the publisher was aware of the falsity if it finds that there were “obvious reasons to doubt” the accuracy of the story, and that the defendant did not act reasonably in dispelling those doubts. It is not, however, the failure to act reasonably in itself that establishes malice; that failure is only a link in the chain of inferences that could (but need not) lead a jury to conclude that the publisher failed to conduct an investigation because it was already pretty much aware of the falsity. As Harte-Hanks points out, “[although failure to investigate will not alone support a finding of actual malice, the purposeful avoidance of the truth is in a different category.” Id. at 692, 109 S.Ct. at 2698 (citation omitted).
A
In opposing summary judgment, Masson introduced a number of items that could sustain a jury verdict that The New Yorker “in fact entertained serious doubts as to the truth of [its] publication.” To begin with, there was the incident involving The New Yorker’s fact-checker, Nancy Franklin. According to Masson, he pointed out to her the inaccuracy of various quotations and “asked permission to review those portions of the article which attributed quotations or information to him, but was brushed off with a never-fulfilled promise to ‘get back to [him].’ ” Masson, 111 S.Ct. at 2424-25. Masson also claims he asked Franklin to speak with Malcolm about the accuracy of the quotations, and that Franklin said she would. Masson Declaration at 7. Again according to Masson, Franklin did assure him that “there were tape-recordings of all the conversations, and they [the quotations] would be verbatim and accurate.” Id. Masson also claims that a number of corrections he suggested to Franklin found their way into galleys but were changed back by the time the magazine was published. Id. at 7-8.4
While the protestations of a source will not, standing alone, support an inference that the publisher “entertained serious doubts” about, or even “had obvious reasons to doubt,” the accuracy of the quotations, see Harte-Hanks, 491 U.S. at 692 n. 37, 109 S.Ct. at 2698 n. 37, the Nancy Franklin incident involves much more. This was not a cold call from a source complaining about the accuracy of a story; Masson’s objection was raised during a fact-checking process initiated by The New Yorker as part of its policy of maintaining the accuracy of its stories. A jury could reasonably conclude that claims of inaccuracy raised in that context would be taken seriously. Franklin’s alleged response— the assurances of accuracy, the representation that she would check with Malcolm, the express or implied promise to verify the quotations against the tapes, and the promise to get back to Masson — plus the fact *901that some of Masson’s suggested changes were made temporarily, further support the inference that The New Yorker did not dismiss Masson’s charges of inaccuracy out of hand. One possible conclusion the jury might draw, were it to credit Masson’s version of the events, is that Franklin did indeed develop a serious doubt about the accuracy of the quotations; that she looked into the matter by talking to Malcolm and/or listening to the tapes; that she (or someone else) decided it was necessary to make some of the changes in order to remove known inaccuracies; and that the quotations were changed back as the result of a conscious decision to sacrifice accuracy to creativity.
But the jury need not go quite so far in its findings in order to hold for Masson. Even if it finds that The New Yorker’s editors did not make a conscious decision to let Malcolm alter Masson’s statements, the jury could nevertheless conclude that The New Yorker had developed “obvious reasons to doubt” the accuracy of the quotations but, in an effort to “purposeful[ly] avoid[ ] ... the truth,” failed to conduct a reasonable investigation of Masson’s claims of inaccuracy.
This conclusion would be supported by other evidence, presented by Masson, suggesting that the magazine allowed Malcolm to edit quotations, much as she edited other portions of her text:
Discussing his plans for Anna Freud’s house, Masson said, “it’s dark and somber and nothing went on in there. Boy, I was going to renovate it and open it up, and the sun would come in and there would be people and — Well, that’s what it needs, but it is an incredible storehouse. I mean the library_” Malcolm’s typed draft reads, “Sun would have come pouring in, people would have come, there would have been parties and laughter and fun.” This sentence is crossed out, and the following sentence, to which Masson takes exception, is handwritten above it: “Maresfield Gardens would have been a center of scholarship, but it would also have been a place of sex, women, fun.”
Masson, 895 F.2d at 1567 (Kozinski, J., dissenting) (record citations omitted). If the jury determines that The New Yorker’s editors were aware of this alteration, it could easily infer that the editors knew Malcolm was fiddling with the quotations, but still decided to shut their ears to Mas-son’s protestations of inaccuracy. All of this would support a conclusion that The New Yorker “in fact entertained serious doubts” as to the accuracy of Malcolm’s quotations, but chose not to investigate the matter.
The New Yorker points to language in Harte-Hanks which says that a publisher’s failure to conduct an investigation to verify the accuracy of reports it obtains from third parties cannot establish actual malice, “even when a reasonably prudent person would have” conducted an investigation. Harte-Hanks, 491 U.S. at 688, 109 S.Ct. at 2696. This is clearly the law; Sullivan itself reversed a libel verdict against a major newspaper even though the paper did nothing to investigate the accuracy of an advertisement that libeled the plaintiff. 376 U.S. at 286-88, 84 S.Ct. at 729-30. What this means is that a publisher who does not already have “obvious reasons to doubt” the accuracy of a story is not required to initiate an investigation that might plant such doubt. Once doubt exists, however, the publisher must act reasonably in dispelling it. Thus, where the publisher undertakes to investigate the accuracy of a story and learns facts casting doubt on the information contained therein, it may not ignore those doubts, even though it had no duty to conduct the investigation in the first place.5
*902Here, there was evidence from which a jury could infer that The New Yorker conducted an investigation into the facts underlying Malcolm’s article, including the quotations, and that this investigation yielded information that gave The New Yorker “obvious reasons to doubt” the accuracy of the disputed quotations. An objective, independent source (the tapes) was readily available to enable The New Yorker’s editors to resolve those doubts, and it would have been prudent for The New Yorker to consult the tapes. See Harte-Hanks, 491 U.S. at 692, 109 S.Ct. at 2698 (newspaper’s failure to listen to tapes contradicting report supports finding of actual malice).6 If The New Yorker believed the conversations in question were not on tape, it could have confronted Malcolm with its doubts and asked her to verify the accuracy of those quotations by producing her notes and other supporting materials.7 To the extent they were actually aware that Malcolm was changing quotations during the editing process, The New Yorker’s editors had a responsibility to ask her to explain a practice that, on its face, is so inconsistent with responsible journalism.
Had The New Yorker produced evidence that it took these steps and then concluded (even unreasonably) that the quotations were accurate, it would have been entitled to summary judgment. But it didn’t. Mas-son has therefore presented a triable issue of fact as to whether The New Yorker “in fact entertained serious doubts” as to the accuracy of the quotations but chose to publish the article anyway. The district court erred in granting the magazine’s motion for summary judgment; the case must go to the jury.
B
Knopf knew that Masson’s lawyer had contacted The New Yorker about the accuracy of the article. Thus, Knopf had “knowledge of at least [Masson’s] general allegation that the article contained defamatory material.” Masson, 111 S.Ct. at 2425. This fact, however, actually weighs in Knopf’s favor. In republishing the article, Knopf was entitled to rely on the investigation of the matter previously conducted by The New Yorker.8 The magazine’s sterling reputation for accuracy and the existence of its fabled fact-checking department gave Knopf sufficient reason for dismissing Masson’s claims that he was misquoted, particularly when he made it clear he had presented the same claims to The New Yorker at the time of the original publication. While it may have been prudent for *903Knopf to investigate Masson’s claims, “[a] ‘reckless disregard’ for the truth ... requires more than a departure from reasonably prudent conduct.” Harte-Hanks, 491 U.S. at 688, 109 S.Ct. at 2696. Thus a publisher who does not have information giving it “obvious reasons to doubt” the accuracy of a story is not required to conduct an investigation that would cause it to develop such a doubt, “even when a reasonably prudent person would have done so.” Id.9
Given everything Masson has alleged Knopf knew at the time it republished Malcolm’s story in book form, we hold that no reasonable jury could find that Knopf “in fact entertained serious doubts” as to the accuracy of the quotations. Under Harte-Hanks it was therefore entitled to summary judgment.
Conclusion
The First Amendment protects the right to debate vigorously the issues of the day, and to avoid stifling that debate we accord wide latitude to authors and publishers. Yet the freedom of the press, like all rights, carries with it responsibilities. One of these is not to abuse the public trust by knowingly or recklessly publishing falsehoods. Janet Malcolm and The New Yorker may have fallen short of this standard. The ultimate resolution of the issue is for a jury after a trial; at this point we hold only that the district court improperly granted summary judgment for these two defendants.
The district court’s grant of summary judgment is AFFIRMED as to Knopf and REVERSED as to Malcolm and The New Yorker; the case is REMANDED for trial.
. The statements were identified by the Court as "Intellectual Gigolo"; “Sex, Women, Fun”; "I Don't Know Why I Put It In”; “Greatest Analyst Who Ever Lived"; and "He Had The Wrong Man." Id. The Court held that as to a sixth statement, "It Sounded Better," "any difference ... is, in context, immaterial" and thus not actionable. Id. at 2436.
. For purposes of this opinion we treat Malcolm as an independent contractor. Masson contends that she was, in fact, an employee of The New Yorker, and that under normal agency principles any dereliction of duty committed by her can be imputed to the magazine. We leave it to the district court on remand to consider Malcolm’s employment status in light of a more fully developed record, and the legal consequences of that status.
. Malcolm contends that some of the disputed quotations came from notes of conversations that were not recorded. Although the procedural posture of the case prevents consideration of this evidence, Malcolm is free to introduce the alleged notes at trial and attempt to persuade the jury that the challenged passages were essentially verbatim renditions of statements made by Masson that were not on tape.
. For example, Masson claims that Franklin asked him if his first wife, Terri, was Polish. He replied: "I said she was not, she was a Jew from Poland, and wanted this distinction made very carefully, since as a child she had been in the Warsaw Ghetto and had suffered from the anti-Semitism of the Poles.” Masson Declaration at 6. In the galleys, Franklin changed ”[s]he is Polish" to ”[s]he is a Polish Jew.” Id., exh. 2; see also exh. 5 (Franklin’s deposition testimony acknowledging making change). On another version of the galleys, Franklin’s addition of “a ... Jew” is accompanied by a query: “Where did this come from?” Id., exh. 3. The published version of the article reads “[s]he’s Polish."
. We are aware that this puts publishers like The New Yorker — whose practice it is to investigate the accuracy of its stories — at something of a disadvantage compared to other publishers such as newspapers and supermarket tabloids that cannot or will not engage in thorough fact-checking. After all, publications that check their stories for accuracy are more likely to develop "obvious reasons to doubt” than ones that do not. That, however, is the standard prescribed by the Supreme Court in Harte-Hanks, and it makes considerable sense: Readers of reputable magazines such as The New Yorker are far more likely to trust the verbatim *902accuracy of the stories they read than are the readers of supermarket tabloids or even daily newspapers, where they understand the inherent limitations in the fact-finding process. The harm inflicted by a misstatement in a publication known for scrupulously investigating the accuracy of its stories can be far more serious than a similar misstatement in a publication known not to do so. This is not to say, of course, that statements in publications less rigorous than The New Yorker cannot be defamatory, or that such publishers will escape liability when they turn a blind eye to known or strongly suspected inaccuracies.
.While the tapes were lengthy, The New Yorker need not have checked every single word in every quotation: If a spot check of several disputed quotations had established that they were essentially verbatim, The New Yorker would have been reasonable in concluding that Mas-son’s complaints of inaccuracy were unfounded. Of course, had a reasonable spot-check revealed material inaccuracies—even non-libelous ones— this should have alerted The New Yorker to conduct a more thorough investigation. In Harte-Hanks, the Supreme Court noted that the tapes of the interview “presented a simple means” of ascertaining the truth. 491 U.S. at 683, 109 S.Ct. at 2693. The Court noted that the "hesitant, inaudible, and sometimes unresponsive and improbable tone" of the source’s voice on the tapes "raise obvious doubts about her veracity.’’ Id. at 691, 109 S.Ct. at 2698. It would be simpler by far to check purportedly verbatim quotations against a tape of the actual conversation, verifying only words rather than tones of voice.
. Where an independent source for verifying the accuracy of quotations or other facts is not readily available, obtaining such verification from an author of good reputation will protect the publisher from liability as a matter of law.
. Masson claims that Knopf, though not The New Yorker, was aware of unspecified problems with one of Malcolm’s prior works. While this factor might be relevant were Knopf attempting to rely on Malcolm’s reputation, we hold that Knopfs reliance on The New Yorker’s reputation is sufficient to shield it from liability.
. Masson claims that the book contains one change from the article. "In her article Malcolm had written about Eissler: ‘whose letters he now returns unread’ whereas in the book she has changed it to 'whose letters he no longer answers.’ ” Masson Declaration at 7. There is no indication that Knopf was at any time involved in the decision to make this or any other change to the article.