Cloning cells or organisms by nuclear transfer is laborious, time-consuming and expensive. The timescale for replication of published results may be on the order of years. But conversely, a successful cloning experiment has the advantage, rare in science, that it can usually be evaluated relatively easily, by DNA testing of tissues collected during the procedure.

Peer review aims to assess credibility but is by and large incapable of detecting dishonesty (although Nature is currently reviewing its procedures with respect to image manipulation). With the caveat that peer review will continue to be based largely on trust, what data can be regarded as sufficient proof of cloning? Before the Hwang scandal, the gold standard for proving that an animal had been cloned from nuclear transfer was to test for identical nuclear DNA fingerprints of the clone and nuclear donor. On the other hand, mitochondrial DNA from the cloned animal should differ from that of the nuclear donor, providing a straightforward way of ruling out sample mishandling or outright fraud.

Keeping in mind that extraordinary claims require extraordinary proof, Nature may in rare cases demand it.

In the light of all that has happened, we think it sensible from now on to ask authors to provide not only nuclear but also mitochondrial DNA fingerprints for all cloning papers submitted to Nature. It should be noted, however, that there may be confounding factors in interpreting such data. For instance, the mitochondrial contribution of the nuclear donor may vary depending on the species, on whether it is an interspecies hybrid, and on the nuclear-transfer technique used. In the case of papers reporting new embryonic stem-cell lines, nuclear DNA fingerprints of the lines should be presented for comparison with existing lines, to help rule out sample mishandling (intentional or accidental) or contamination with other cell lines.

How much data should be provided when papers are submitted? Authors of cloning papers should always present enough data to document the logical flow and efficiency of a cloning procedure. But we may in addition require authors to provide raw data on request, for inspection by reviewers or editors. This allows an additional level of verification should questions arise during the review process, by ensuring that the data presented in the paper are an accurate interpretation of the raw data.

Independent tests

In the light of the extraordinary circumstances surrounding the Hwang case, Nature commissioned an independent scientist to verify that the dog Snuppy (B. C. Lee et al. Nature 436, 641; 2005) is indeed a clone, by DNA fingerprinting analysis of blood samples from Snuppy and the nuclear donor. The indications are positive and the results are being peer-reviewed as we go to press. However, a few scientists have suggested that Nature should make such independent tests a condition of publishing cloning papers.

After weighing this suggestion carefully, we concluded that imposing such a standard on the cloning field as a condition of publication would be an overreaction, and one with myriad inherent logistical problems. For example, who will pay for testing, how will mishandling of samples be prevented, and how will the scientists running the verification tests be acknowledged for what could be significant additional work? Moreover, because the cloning field is not unique in its susceptibility to fraud, it would follow that we should require independent verification of all our papers, which is untenable in the current system. The gains of detecting the rare cases of fraud would be negated by the impediments to publication this would bring.

However, in the best interests of science, we encourage researchers embarking on landmark cloning studies to seek independent verification themselves, and to include a report of these findings in their initial submission. And keeping in mind the principle that extraordinary claims require extraordinary proof, Nature may in rare cases demand it.

Nature and most other journals require as a condition of publication that authors make relevant reagents available to the scientific community, and encourage the deposition of cell lines and mutants in established repositories. But it has now been suggested that journals take this a step further, and require the deposition of critical samples in repositories such as the American Type Culture Collection, where they can be cheaply and rapidly made available to other researchers following publication. We are currently weighing up the practicalities of implementing such a requirement, and we would welcome feedback on whether we should and, if so, how to best do it.

Regardless of what repository is used, we urge scientists embarking on what are likely to be landmark cloning studies to ensure that critical samples are properly stored for later verification. As part of this procedure, an independent scientist not involved in the study should obtain and store cells from the nuclear donor, oocyte donor and the resulting animal or stem-cell line (or oversee their deposition in a repository). This precaution is especially important in the case of human donors, where it may not be possible to go back to the subjects to obtain additional tissues for later verification. Funding agencies should make granting dependent on procedures to ensure later verification of samples, and institutions should demand this for approval by the institutional review board.

The Hwang debacle reminds us that science is largely a self-correcting process in which scientists, editors, reviewers, journalists, funding agencies and institutions all play crucial corrective roles. In the aftermath of this deception, we should all undertake close scrutiny of our procedures and standards, with an eye towards preventing it from ever happening again. We should be vigilant against knee-jerk reactions and witch hunts. Nature welcomes feedback on its approach (e-mails may be sent to authors@nature.com).