delivered the opinion of the court.

The prisoner was indicted in the county of Dekalb, for the crime of murder, convicted and sentenced to death, and he prosecutes an appeal to this court.

The question of reversal rests upon the character of the proof adduced on the trial and the conduct of the jury in the mode adopted by them of arriving at the result of this verdict as it is exhibited in affidavits of three of the jurors.

The testimony is most of it entirely circumstantial, perhaps all of it, except a portion of negro Linda’s, who says the prisoner told her he had shot the deceased, which we think, upon the whole view of the case, entitled to but little weight.

It appears, that the prisoner and a negro named George, the husband of witness Linda, were lurking in the neighborhood, both armed, both having threatened to kill the deceased, who they suspected of being employed to arrest them. On the night the murder was committed, Linda swears, that the prisoner came to the place of her residence early; stayed there till bed time; she asked him if he would lie down, which he declined; that she went to bed and left him there; that she waked several times during the night and always saw him.

In addition, two other witnesses, Beck and Nancy, both belonging to the master of Linda, swear positively that the prison*291er was in the house of Linda at the time the gun was fired that killed deceased; that they heard the gun, and Beck says, that she afterwards heard prisoner talking with Nancy in Linda’s house; and Nancy says, that after the gun was fired she went into Linda’s house, found prisoner asleep, waked him, and conversed with him: powerful proof; conclusive proof, if true. But the jury chose to believe Linda, and to disbelieve Beck and Nancy. If they had done this upon a proper exercise of their judgment, the question would have been presented in a very different point of view from what it is; but this was not done. Instead of weighing the credibility of these witnesses by the legitimate mode of general weight of character and probability of their statements, they have chosen to shut themseves in their room, send out the .constable, talk to each other in a louder tone than common, and then enquire of him if he heard them; and upon his reply in the negative, they have considered it as conclusive, that Beck and Nancy could not have heard the prisoner talking in Linda’s house. This mode of arriving at the truth of testimony cannot be permitted; it is too vague and uncertain. A different intonation of voice, a difference in the structure of the rooms, would destroy its virtue as a test; and besides, they had to take the word of the constable as to the fact whether they were heard.

The circumstances of the case, with the exception of the tracks from the place where this murder was committed and the shoe of the prisoner, which was produced on the trial, in comparison therewith, apply nearly, if not quite as well, to George as the prisoner. Both were apprehending that the deceased had been employed to arrest them; both had threatened to kill him; both were in the neighborhood, and had plenty of opportunities to carry their designs into effect.

The shoe which was produced on the trial against the prisoner, was half an inch longer than the tracks: this was another difficulty the jury found much in their way; but supposed that in running, a track might be shorter than the shoe: they and their constable again try an experiment by running, and find their own tracks half an inch shorter than their shoes; and a verdict of guilty is returned.

*292Independent, then, of the doubtful and uncertain character of the proof introduced to establish the guilt of the prisoner, we cannot permit verdicts which have been obtained like this, upon uncertain and dangerous experiments, instead of a calm, deliberate and philosophical examination of the proof, to stand where the lives of individuals are at stake.

The judgment is therefore reversed, and the case remanded for a new trial.

See Kirby vs. The State, 3 Humphreys.