delivered the opinion of the court. The first question raised on the argument, is, whether error in fact can be assigned, on a writ of error, coram vobis. It was contended, that error in fact is only to be corrected in the same court where the cause was commenced, by a writ of error coram nobis. This is certainly against what has been the universal practice with us, as to writs of error to inferior courts, and we apprehend the objection has no foundation in principle. It is true, we find it laid down in the English books, that for error in fact, a writ of error will not lie from the exchequer chamber, or the house of lords, but the reason assigned for it shows that no such rule can apply to this court. Error in fact must be tried by a jury, and no such trial can take place, either in the house of lords, or in the exchequer chamber. Hence the necessity of a writ of error coram nobis. Although in England; error in fact may be tried on a writ of error coram nobis, in the common bench, yet, we find that error in fact is assigned, sometimes, on a writ of error from K. B. to the C, B. The case of Carre v. Barker, (Cro. J. 250.) is one of this description> and infancy was the error there assigned. This court can issue a venire for a jury to try the fact of infancy. There can, therefore, be no reason why this error should not be corrected here.
It is said, however, that the error is not well assigned in this case, because it only avers the infancy at the time of appearance and plea pleaded; whereas the averment should have been, that he was an infant at the time of the rendition of the judgment. This position does not seem to be supported by any adjudged cases; and it is not warranted by the precedents. (2 Lil.Ent. 490.) The reason urged why the infancy should be referred to the time of the rendition of the judgment is, because the appearance and defending the cause, after the party attained his full age, would be a waiver of the error. Admitting that such appearance and defending the action would be a waiver, that could not be taken advantage of on this demurrer, unless *428the fact appear from the record, and, therefore, admitted by the demurrer. No such fact appears. The time when the judgment roll was made up and filed cannot, certainly, affect the question as to a waiver. It must, if at all available, be the appearance and defending the action at the trial; and, at that time, Duncan had not attained his full age, as is to be collected from the record. He was twenty years, four months, and twenty days old, on the third Monday in August, 1815 ; the trial appears to have been in December following, and the rule for jugdment must have been entered at that time, according to the course and practice of the courts of common pleas.
In assigning the infancy as error, there may, perhaps, be an informality, in not alleging that Duncan was no more than of the age set forth. This, by the precedent before referred to from Lilly, appears to be the form. But I apprehend this is an informality that would, at all events, require a special demurrer. The assignment of error alleges that he was, at the time of appearance and pleading, under the age of twenty-one years, to wit, of the age of twenty years, four months, and twenty days. This would seem to contain, in substance, every thing that was material and necessary.
There is no foundation for the objection, that the writ of error was sued out before the judgment was rendered in the court below.The rule for judgment must have been in December, 1815; and the writ of error was not sued out until the January following. A writ of error may be tested before judgment is given. It is sufficient if the judgment be given before the return of the writ. This is the usual course for preventing execution. (2 Tidd’s P. 1062. 1 Johns. Rep. 493.) It is not the signing of the judgment roll, but the rendition of the judgment which forms the test. The signing is the mere authentication of the judgment. In the case of Jaques v. Nixon (1 Term Rep. 280.) it is admitted to be the settled practice to sue out a writ of error before judgment is actually signed. It cannot be true, in fact; as was' suggested on the argument, that the writ of error was returned and filed before the judgment in the court below was actually signed; at least, that is not to be intended fronj any thing appearing on the demurrer book. The judgment) it is true, appears to have been signed in July, 1816, and the writ of error to be returned in the preceding May term of this court. But if the return, in point of fact, was made in the May vacation, it would relate back to the term-; and it cannot be intended that the *429record was signed after the writ of error was, in fact, returned and filed in this court, with the record of judgment in th~ court below annexed to it.
We cannot reverse the judgment as to Duncan, and affirm it as to the others. Where the judgments are distinct, we may in part, and affirm in part, as in cases of damages and costs; but when the judgment is entire, there must be a total affirmance or reversal. This point was settled in this court, in the case of Richard v. Walton, (12 Johns. Rep. 434.)
According to the case of Dewitt v. Post, (11 Johns. Rep. 460.) judgment of reversal, for error in fact, is revoeetur. The defendant may, therefore., if be chooses, have leave to his demurrer, and rejoin to the assignment of errors.
Judgment accordingly, with leave, &c.
Dewitt v. Port, 11 Johns. Rep. 460. Tidd's Pr. 1058, 1057. 1126. 2 Bac br. 503. 2 Comyn's Rep. 600 2 Crompton's Pr 395 2 Sellon's Pr. 531. 1 Str. 127. 607.