Three of six defendants in the same indictment were tried for burglary. The witnesses to connect them with the offense were two of those not on trial, which two were charged as principals in the second degree. They testified positively to the participation of the others in the burglary; and while they were fully corroborated by other witnesses as to the fact that a burglary was committed, and as to some of the circumstances that attended it, there was no corroboration of their *315testimony as to the fact that their three co-defendants then on trial, were concerned in it. No circumstance, whatever, implicating these three, or any one of them, was otherwise proved. There was nothing else to cast the slightest suspicion upon them. All the goods missed from the house by the owner were found in the possession of that one of the defendants who was neither upon trial nor a witness.

In giving their evidence, the two accomplices avowed their own presence, and admitted that they kept watch while the others broke the house, entered it, and brought out the goods. They gave, substantially, the same account of the whole criminal transaction,.but differed from each other in several particulars connected with other incidents and occurrences of the night whilst they were together.

Without going into the question discussed in Childers vs. The State, 52 Georgia Reports, 106, touching the difference between corroboration generally and corroboration upon the particular fact on which the whole pressure of the case rests— a question which divided the court as then constituted — we are of opinion that, in the present case, the inherent weakness of accomplice-testimony was made still more feeble by contradictions between the two witnesses as to minor matters, in which, if they had been truthful, they ought to have agreed. As a specimen of their divergence a single instance may be given, among several, which the record exhibits: After the burglary the witnesses proceeded on their travels to get a dog. One of them says they got the dog about midnight and started back; that when they started back with the dog they thought they would take a hunt, but the dog ran away from them, and they got lost and made up a fire, and staid in the woods until about ten o’clock next day. The other says they did not get the dog until next morning, and did not start back that night nor go hunting. Agreement between the two as to all the details of the burglary, and the want of such agreement as to the subsequent occurrences of the night, would seem to give room for thinking that these persons possessed inventive faculties and are not indisposed to exercise *316them. But the prisoners on trial defended by setting up an alibi. They proved themselves to have been at home, two of them by one witness, with some appearance of corroboration by another, and one of them by four witnesses. There was also evidence that this last was a man of good character. The trial took place within a week after the burglary, and on that account there was less than the usual danger of mistake by the witnesses as to time. Allowing for every uncertainty in that respect, we think the case made out by the state through the tainted evidence of avowed accomplices was met and overcome. The verdict was contrary to the evidence, and the judgment of the court refusing a new trial is, on that ground, reversed.

Judgment reversed.