Reference for Pesachim 24:1
אלא בין יממא לליליא לא טעו אינשי ובדין הוא דניתיב לקמיה טפי אלא שבה' חמה במזרח ובז' חמה במערב
but that people do not err between day and night; and logically we should give them more at the end,<span class="x" onmousemove="('comment',' Lit., 'forward'.');"><sup>1</sup></span> but that in the fifth hour the sun is in the east while in the seventh the sun is in the west.<span class="x" onmousemove="('comment',' According to R. Meir: if A testified that the crime was committed in the second hour, and B that it took place in the third, their testimony is valid (v. Supra 11b) , unless they are rebutted over the whole period in which an error is possible. Thus A, if rebutted, can plead that he erred, and that the crime actually took place either in the first hour or in the third or fourth. He should also be able to plead that it took place within the hour before sunrise, since R. Meir allows for an error of nearly two hours, but that he would never mistake night for day. Similarly B, if rebutted, can plead that he erred, and that the crime took place at any time between the first and the fifth hour. Hence they are liable to be rebutted over the whole of this time; i.e., C and D testify that they were elsewhere from the first until the fifth hour, and such rebuttal is designated a rebuttal in respect of hours, and therefore the evidence, if unrefuted, is valid. By the same reasoning, according to R. Judah, who allows for a margin of nearly three hours' error, the period is from the first until the sixth hour, the seventh being disregarded, as explained in the GEMARA: - This wide latitude is granted only in so far that the witnesses will not be subject to retaliation (v. p. 53, n. 4) otherwise, but the evidence none the less may be void. E.g., if it is necessary to assume that B erred in two hours and that he really meant the fifth hour, A's testimony cannot be reconciled with it on any reasoning, and as we are left with one witness only the accused cannot be condemned.');"><sup>2</sup></span>