Peer review is also a device by which entire scientific thoughts can be excluded.
If the proponents of theory 'A' are in a position to and decide to exclude from publication journal articles which indicate data support theory 'B' and refute theory 'A', those articles do not get published.
A colleague wrote a paper on paleopatterned ground on White Butte in North Dakota, and his field work indicated that there were indeed permafrost polygons there on the butte, a remnant of the last ice age.
Despite the evidence, the paper was rejected because "...there never has been any permafrost in North Dakota".
Circular reasoning is as pretty as such rejections of valid findings get.
When one considers the reputations on the line and the vast sums of ongoing grant money for findings which enable the manipulation of the entire global economy, mere logical error would seem benign by comparison to the motivation present to stifle debate.
I share your concern. Even if there was no permafrost on the bulk of level ground in North Dakota, why might there not have been on the higher elevation microclimate of a butte. Incidentally, do you know how high this butte is, and can your scientist friend show sufficient variability between the lower ground temperature and the butte surface temperature to include in a resubmission of his article with this argument? At least such modern temperature data could make a good letter to the editor argument.