CARDOZO, J.

The body of Francesco Manzella was found on October 30, 1911, at the bottom of a deep ravine along the Webster road near the city of Rochester. The head and legs had been cut off, and the body forced into a barrel. There must have been a violent struggle before death. The coroner's physician describes twenty-two wounds and bruises. Wounds on the temple had made the victim unconscious; and then with a sharp knife the head had been severed. Whoever used the knife disclosed a surgeon's skill. The legs were cut off later; and again a surgeon's skill was shown. Manzella was a strong, athletic man; and the wounds and bruises bear witness to a vigorous resistance. The defendant is a cripple; he has lost both his legs. He was under arrest within thirty-six hours of the murder; he did not show a scratch or a blood stain. The People connect him with the murder by evidence that he attempted to secrete the body. They connect him in no other way. We are to determine whether the evidence sustains a verdict that he was a principal in the crime.

There was a joint indictment of Domenico Galbo, the defendant, and his brother, Guieseppe or Joseph, but Domenico was tried alone. The brothers were in business together. Their business was the sale of bananas. They had a store on the corner of Railroad street and the Public Market in Rochester; and in the rear was a barn where they kept their horses. Domenico was unmarried, and slept above the store. Joseph was married, and lived at the home of his father-in-law. They had some acquaintance with Manzella, who slept once, if not oftener, above the store with Domenico. Manzella had served a term in prison for extortion. He was known as a "blackhander." It seems to be conceded that he made a business of blackmail. On Saturday, October 28, 1911, two days before the finding of his dead body, he requested a loan of Joseph Galbo, but was repulsed. He then said that he would get the money from Joseph's father-in-law, one Ollis. There is evidence that later in the day he made a like request of Ollis, and was told to come back in the evening. He was never seen again. On Monday morning, before it was yet dawn, the farmers and marketmen driving with their burdens toward Rochester, along the Webster road, saw the Galbo wagon, drawn by a gray and a sorrel horse, and Domenico Galbo in the seat. It was going away from the city. In the rear there seemed to be a barrel covered with canvas. Some of the witnesses could describe the wagon and the horses only. Others had noticed the barrel. Others, though unacquainted with the driver, recalled his appearance. An hour or two later the same wagon with the same driver came back along the same road. Those who saw it then make no mention of a barrel. Two men identify the driver as Domenico. One of them jumped on the wagon and rode part of the way.

By eight o'clock on that morning the body of Francesco Manzella was at the bottom of the ravine. It had come there since the afternoon of the day before. On Sunday, October 29, about three o'clock in the afternoon, a young man went to the ravine to set a trap for a skunk. There was no barrel and no body then. Eight o'clock the next morning the same man went to the ravine, found the barrel and the dismembered body, and notified the police. On the afternoon of the same day the Galbo brothers were in jail.

A trail of circumstantial evidence leads from the ravine to the Galbo store. The barrel was a wine barrel with iron hoops; in the Galbo store four barrels of the same kind were found by the police. Adhering to the barrel in the ravine was a fragment of a printed card. The card of one Blandi, a wine dealer of Pittsburg, was identical in size and form and space of type. Five barrels of Blandi's wine were sold in March, 1910, to Joseph Galbo. The waybill and receipt produced by the railroad company establish its delivery. Five barrels reached the Galbo store; four were still there after the body was discovered. Near the body in the ravine was a printed time card. The cards had been printed for one Hahn of the McCabe Electrical Company, who kept them in his desks. The desks were sold to the Capon-Sullivan Company, which occupied a store owned by Joseph Galbo. It was next door to the store of the Galbo Brothers. Less than a week before the murder, on October 23, 1911, a member of the Capon-Sullivan Company found the time cards in the desks and threw them into the Galbo yard. As late as November 1 some of the cards were discovered in front of the Galbo barn. Near the barrel in the ravine there was a part of a burlap sack, which had once been filled with chicken feed. It was stained with blood, and had doubtless been used to cover the top of the barrel. It was stamped with the label of the Dickinson Company of Chicago. It bore the tag of "Lathrop's Pet Shop." Six bags of the same kind, with the same label, were found in the Galbo store, and there was another in the Galbo wagon. A fortnight before the murder, Joseph Galbo bought from nine to twelve sacks of chicken feed from the Clark Douglass Company, which bought them from "Lathrop's Pet Shop." Near the burlap sack were also parts of a rope. Rope of the same material and weight was suspended from hooks, and carried bunches of bananas in the Galbo store. The rope from two hooks was missing. The tailboard of the wagon showed traces of white paint. White paint was found on the fence which ran along the Webster road at the top of the ravine. Discolored shavings of wood were found in the barrel, and other shavings, apparently discolored in the same way, were found in the Galbo barn. A chemist showed that all the shavings had been colored by the same dye. In the stall of the barn there were breaks in the cement which indicated a recent excavation of a size suitable for a grave. But nothing else that even remotely suggested guilt was found either in the barn or in the store. The police took possession at once; ransacked the buildings from top to bottom; tore the woodwork open and searched in every nook for traces of blood and for implements of crime. Nothing was found. A detective was then stationed in a nearby cell to listen to the brothers' talk. Domenico said: "They are looking for the driver of the wagon." "You drove the wagon," said Joseph. "I know I did," said Domenico.

This in rough outline was the People's case. The defendant met it with a sweeping denial. He denied that he had driven a wagon along the Webster road on the morning of October 30. He often went along that road with his burden of bananas, but he did not go that day. That day he drove to Fairport, in a different direction. He did not leave the barn till about six o'clock, and when he left, he did not take the gray horse along. The gray horse was sick, and could not go. That is the defendant's statement. But a witness for the People, one Cira, who worked in the Galbo barn, said that he reached there before six, and that Domenico and the gray horse were gone. It would not be helpful to review the defendant's narrative at length. The jury found him guilty of murder in the second degree, and to reach that verdict must have found that his narrative was false. The credibility of witnesses is not for our consideration except where the judgment is of death. We must, therefore, assume that the defendant did drive along the Webster road in the early morning of October 30, 1911. He was there, and spoke falsely when he denied it; he was bearing with him some burden, which to onlookers seemed to be a barrel; and about that hour a barrel, proved by many tokens to have come from the Galbo store, landed with the dead body of Francesco Manzella at the bottom of the ravine. The jury had the right to find that Domenico Galbo had the body with him, and threw it into the ravine to bury it from the sight of men.

The People say that these acts of possession and concealment stamp the defendant as the murderer. They do, we think, beyond question justify the inference that in some way and at some stage he became connected with this crime. But the question remains, in what way and at what stage? Was he a principal, and if so, did he himself commit the offense, or aid and abet its commission, or counsel or induce another to commit it? Was he, on the other hand, an accessory after the fact, aiding the offender to avoid arrest or punishment? Principals in the first and second degree at common law, and accessories before the fact (Whart. Cr. L. [11th ed.] §§ 240, 245, 263), are classed alike as principals to-day (Penal Law, § 2). Accessories after the fact are classed simply as accessories. Which of these degrees of guilt attaches to the defendant?

It is the law that recent and exclusive possession of the fruits of crime, if unexplained or falsely explained, will justify the inference that the possessor is the criminal. That rule has most frequently been applied in cases of burglary ( Knickerbocker v. People, 43 N.Y. 177) and larceny ( Stover v. People, 56 N.Y. 315) and receiving stolen goods ( Goldstein v. People, 82 N.Y. 231); but it is not unknown in cases of murder ( People v. Jackson, 182 N.Y. 66, 78; Wilson v. U.S., 162 U.S. 613, 619; Williams v. Comm., 29 Pa. St. 102, 106). The highway-man kills his victim; the purpose of the murder is robbery; the same inference that identifies the robber identifies the murderer. Possession of the dead body — the subject of the crime itself — has much the same significance as possession of jewels or money or other fruits of crime. If there is any distinction, it is one chiefly of degree. The fruits of crime are themselves objects of desire; the possessor, at least presumably, has them because he wishes to enjoy them. But the possessor of the dead body wishes only to be rid of it. Its possession is thus associated more readily than that of money or jewels with the notion of concealment and thus with the form of guilt that attaches distinctively to the accessory after the fact. Often an attempt to secrete the body has played no other part than that of corroborative evidence ( State v. Dickson, 78 Mo. 438; People v. Jackson, supra; Burrill Circ. Ev. p. 440). We do not say that it may not sometimes, if not explained or rebutted, be sufficient by itself. We must look to all the circumstances.

Only half of the problem, however, has been solved when guilty possession fixes the identity of the offender. There remains the question of the nature of his offense. Here again the facts must shape the inference. Is the guilty possessor the thief, or is he a receiver of stolen goods? Judges have said that if nothing more is shown, we may take him to be the thief ( R. v. Langmead, 9 Cox C.C. 464). But as soon as evidence is offered that the theft was committed by some one else, the inference changes, and he becomes a receiver of stolen goods ( Goldstein v. People, supra; People v. Friedman, 149 App. Div. 873, 877; 2 Russell on Crimes [6th ed.], p. 287 et seq.). Sometimes the circumstances may make it proper for a jury to say which inference is the true one ( R. v. Langmead, supra). Learned commentators have said that in many cases the wrong inference has been drawn (Best on Ev. § 210; Burrill Circ. Ev. 456; Wills Circ. Ev. [6th ed.] 91, 92). Men whose crime was that they had received stolen property, have been convicted of stealing it themselves (Best, supra). A warning was sounded as long ago as the time of Lord Hale. These presumptive evidences, he said, "must be very warily pressed, for it is better five guilty persons should escape unpunished, than one innocent person should die" (2 Hale P.C. 289). He couples the warning with instances drawn from his own experience where wrong had been done.

The problem is a hard one. To solve it we must steadily bear in mind that the inference of guilt to be drawn from possession is never one of law. It is an inference of fact (Wigmore on Ev. § 2513). Other facts may neutralize it, or repel it, or render it so remote or tenuous or uncertain that in a given case we should reject it. The man who secretes a body and lies about it, may be found in most cases to be concealing his own crime, and, therefore, to be the murderer. That is so because personal guilt, unless the circumstances point to some other connection, is the reasonable inference. We are not to assume without evidence that some one else is implicated. He who conceals the crime may be taken to be the perpetrator. But how, if he proves an alibi? Are we then at liberty to infer that even if he did not commit the murder himself, he incited some one to do it, and thus, in spite of his proved absence, hold him as a principal? A is seen to shoot B, but C later has the body, and will not tell how he came by it. The law must say whether his silence is to condemn him as principal or as accessory. The same problem arises if a child or a frail woman secretes the body of a man who has been robbed and murdered on the highway. In these cases, the inference of actual perpetration is repelled. Even then, incriminating inferences remain possible; but unless other circumstances are shown, there is no principle of selection, aside from the presumption of innocence, to guide the choice between them. The guilty possessor of the body, though he did not use the weapon, may still have aided and abetted; but unless there are tokens that several joined in the affray, the likelihood of his presence is no greater than the likelihood of his absence. He may still be an accessory; whether before the fact or after is the problem. If the circumstances make one inference just as reasonable as the other, we must give the defendant the benefit of the conclusion that would mitigate his guilt ( People v. Lamb, [Ct. of App.] 2 Abb. Pr. [N.S.] 148, 165).

In this case a legless cripple is charged to have murdered a strong man. The murder followed a fierce fight in which the strong man was beaten and wounded. It seems certain that the wounds were inflicted and the head severed as parts of a single combat. We cannot with reason say that the cripple did these things. Least of all can we say that he was able to do them and escape without a scratch or a blood stain. "Insufficient evidence is, in the eye of the law, no evidence" ( Matter of Case, 214 N.Y. 199, 203); and when we say that something is impossible, we do not mean impossible in the strictest sense, but so nearly impossible that a jury ought not to believe it ( Hudson v. Rome, W. O.R.R. Co., 145 N.Y. 408; Matter of Harriot, 145 N.Y. 540, 545). But as soon as we concede that the defendant did not kill Manzella, we lose ourselves in mystery when we attempt to measure the degree of his connection with the crime. We have no evidence, direct or circumstantial, that the actual perpetrator was assisted by any one. We have nothing to tell us when or where the crime occurred. We have no sign that it was committed in the defendant's presence. He may have known of it in advance, and planned or encouraged it. He may have learned of it later, and attempted to shield the criminal. The trial judge told the jury that the burden was on the People to prove beyond a reasonable doubt that the defendant, though he did not kill with his own hand, was none the less a principal; he must have become connected with the crime while Manzella was yet alive. If all that he did was to help the murderer to escape, he was not a principal, but an accessory, and the jury under the charge were then required to acquit him (Penal Law, § 2). The charge is sound, but it propounded to the jury a problem incapable of reasoned solution. Men do become accessories after the fact, and help others to cheat the law. Particularly is that so when they are related to the perpetrator. We have had to deal with such cases in this court. Such a case was People v. Farmer ( 196 N.Y. 65) where a husband tried to shield his wife by burying the victim's body. That the defendant has lied about the crime does not prove that he must have been implicated as a principal. There is a motive to falsify, whatever the degree of the connection. An accessory after the fact may be punished by imprisonment for not more than five years, or by a fine of not more than $500, or by both (Penal Law, § 1934). There is thus a serious penalty to be avoided. Moreover, the very fact that one becomes an accessory is proof of the strength of the desire to shield the principal.

In these circumstances we cannot see that the jury had any chart or compass by which to guide their judgment. A conviction upon circumstantial evidence is not to be sustained unless the circumstances are inconsistent with innocence ( People v. Harris, 136 N.Y. 423, 453; Lopez v. Campbell, 163 N.Y. 340, 347). We may multiply inferences at times, but in multiplying them, we must not refine and rarefy them beyond measure. A body is hidden. The evidence forbids the inference that the hider is the slayer. That inference excluded, something more must be shown, some probability of time or place or circumstance, before the concealment of the body can be said to prove anything more than concealment of a crime. Small things may turn the scale. But something there must be.

We are thus led to the conclusion that the defendant was not proved to be a principal in the commission of this crime. He ought to have been indicted, and might then have been convicted, as an accessory (Penal Law, § 2). It is a mere accident that the jury found him guilty of murder in the second degree. To sustain their verdict we must be prepared to hold that on the same evidence they might have sent him to his death. The law is not so lax in its forfeiture of life. We cannot close this record with any sense of assurance that the defendant's connection with the crime preceded the event. The People charge in the indictment that the brother, Joseph, was one of the murderers. Like the defendant, he did not show a scratch or a blood stain. If the crime was his work, the defendant had a strong motive for concealment. If it was another's work, the mystery is deepened. We know that Manzella had led a life of crime in which bitter enmities must have been aroused; and we cannot say how the defendant was connected with the man or men by whom those enmities were avenged. In connecting him as a principal, conjecture has filled the gaps left open by the evidence, and the presumption of innocence has yielded to a presumption of guilt.

The judgment of conviction should be reversed, and a new trial ordered.

WILLARD BARTLETT, Ch. J., HISCOCK, CHASE, COLLIN and SEABURY, JJ., concur; HOGAN, J., concurs in result.

Judgment of conviction reversed, etc.