By the Court,
Garber J.:
The appellant was convicted of an attempt to commit arson. A motion for a new trial was overruled. The testimony was circum*341stantial. The Court below, at the request of the prosecution, gave to the jury the following instruction : “ If the testimony establishes the guilt of the defendant beyond a reasonable doubt, it is your duty to find him guilty, although the testimony is but circumstantial. The term ‘ reasonable doubt,’ does not mean a mere possible doubt, because everything relating to human affairs and depending on moral evidence, is open to some possible orjmaginary doubt. A reasonable doubt is that state of the case which, after comparison and consideration of all the evidence, leaves the minds of the jurors in that condition that they cannot say they feel an abiding conviction to a moral certainty of the truth of the charge. The jury must be satisfied from the evidence of the guilt of the defendant ,beyond a reasonable doubt, before they can find him guilty; but in order to justify the jury in finding the defendant guilty, it is not necessary that the jury should be satisfied from the evidence of his guilt beyond the possibility of a doubt. Although the jury may think that it is possible that the defendant did not commit the crime, yet if the jury is satisfied from the evidence of the guilt of the defendant beyond a reasonable doubt, they are bound to find him guilty, whether that guilt is so proved by direct or circumstantial evidence. If jurors were to disregard circumstantial evidence, there would be an end to the administration of law and of government. Although there have been rare cases where innocent persons have been found guilty upon circumstantial testimony, yet it must be remembered that notwithstanding all that can be urged against it, this kind of testimony is, in the judgment of those most experienced in the investigation of truth and the administration of justice, not unfre-quently as satisfactory, if not more so, than the positive testimony of individuals; and the jury should not attach too much importance to the teachings of such cases. The eye of omniscience can alone see the truth in all cases; circumstantial evidence is there out of the question ; but clothed as we are with the infirmities of human nature, how are we to get at the truth without the concatenation of circumstances ? Though in human judicature, imperfect as it must necessarily be, it sometimes happens that in a few instances, owing to the minute and curious circumstances which sometimes envelope human transactions, error has been committed, yet this species of *342evidence is more satisfactory than the testimony of a single individual who swears he has seen a fact committed.”
The concluding portion of the instruction lays it down as a rule of law, that circumstantial evidence is more satisfactory than direct, where the latter consists of the testimony of a single eye-witness : that is, that circumstantial testimony is more satisfactory than another given species of testimony which may be sufficient to warrant a conviction. (Comm. v. Tuttle, 12 Cush. [Mass.] 504.)
It is contended that the instruction refers to circumstantial evidence in general and as a species, and not to the absolute or comparative weight of the evidence in this particular case; and this seems to be the construction most favorable to the correctness of the instruction. To have instructed the jury that the evidence before them was more satisfactory than would be that of a single eyewitness, would have been manifestly erroneous — a clear invasion by the Court of the province of the jury, who are the sole and exclusive judges of the “ sufficiency in fact ” of the evidence. It would have been deciding a question of fact,' not laying down a rule of law.
If the instruction is that circumstantial evidence is universally and always more satisfactory than direct evidence, or than a specified mode of direct evidence, is it not equally erroneous ? Is there, or can there be such a rule of law ? Suppose one biased witness swears to each of a series of facts, from which, if believed, an inference of all the facts constituting guilt can fairly be drawn. Would such circumstantial evidence be deemed more satisfactory than if one fair witness had sworn directly to the facts inferred ? Circumstantial evidence may or may not be more satisfactory than the testimony of an eye-witness. Whether it is or is’not, depends upon the circumstances of each particular case. The integrity, capacity and means of knowledge of the witness who testifies directly on the one hand; and the integrity, capacity and means of knowledge of the witness or witnesses who testify to facts from which other facts are to be inferred, together with the correctness of the inferences drawn, on the other hand. Thus depending, the question is one of fact, not of law. The law cannot declare in general which is the more satisfactory, by any defined combinations *343of facts, so much does the question depend upon the minute and peculiar circumstances incident to each case. (Starkie Ev. 411.) The same great writer further says: “ In the abstract, and in the absence of all conflict and opposition between them, the two modes of evidence do not in strictness admit of comparison. In theory, circumstantial evidence is stronger than positive and direct evidence, wherever the aggregate of doubt arising first upon the question, whether the facts upon which the inference is founded are sufficiently established; and secondly upon the question whether, assuming the facts to be fully established, the conclusion is correctly drawn from, them, is less than the doubt: whether in the case of direct and positive evidence, the witnesses are entirely faithworthy.” '(Ib. 525-6.)
We find the same doctrine in other standard works. “ There is no necessity for raising or rating circumstantial evidence, in general, higher than direct. In many individual instances it may be superior in proving force to other individual cases of proof by direct evidence. A chain of circumstances, each proved by eye- or ear-witnesses, each capable of being contradicted or disproved, all submitted to the plain sense of a jury of intelligent men, is often more to be relied upon than the direct and positive assertions of a witness who may not be intelligent or who may be dishonest. But a judgment based upon circumstantial evidence cannot, in any case, be more satisfactory than where the same result is produced by direct evidence free from suspicion of bias or mistake. * * * In truth, these two kinds of evidence ought not, in a general view of their merits, to be contrasted or set in opposition. * * As to the argument founded upon the abundance of circumstances, opportunities of contradiction they afford, &c., while each of these adds greatly to the probative force of circumstantial evidence in particular eases, they have clearly no connection with the value of circumstantial evidence in the abstract.” (Burrill Circ. Ev. 229-236; Wills C. Ev. 29-43.)
The instruction cannot be sustained, as comparing circumstantial evidence in the abstract, either with direct evidence or the direct evidence of one witness. Chief Justice Gibson says that circumstantial evidence is, in the abstract, nearly though perhaps not quite *344as strong as positive evidence: Wills, that “ The best writers, ancient and modern, on the subject of evidence, have concurred in treating circumstantial as inferior in cogency and effect to direct evidence — a conclusion which seems to follow necessarily from the very nature of the different kinds of evidence” : and Phillips, that “ Circumstantial evidence has, in some instances, undoubtedly been found to produce a much stronger assurance of the prisoner’s guilt, than could have been produced by the most direct and positive testimony. As a general principle, however, it is certainly true, that positive evidence of a fact, from credible eye-witnesses,- is the most satisfactory that can be produced, and the universal feeling of mankind leans to this species of evidence, in preference to that which is merely circumstantial.”
On the other hand, Mr. Justice Park, if he is correctly reported, in charging a jury used the language of the concluding portion of this instruction, with the exception that he said “ much more satisfactory,” instead of “ more satisfactory.” We have seen no authorized report of the case (Rex v. Thartell) in which this charge is said to have been delivered; but an extract from it is copied into the text books, and has received the sanction of a dictum of the Supreme Court of California.
This extract evidently comprises but a small portion of the charge, and the omitted portion may have so guarded and qualified the expression as to bring it in harmony with the other authorities. "Very likely, the expression actually used was, “ often much more satisfactory,” as we find it generally so used by the English judges in their charges and opinions. Wharton cites Wills as authority for the extract. The case is cited by both Wills and Burrill, but in another connection-; and although they devote a considerable portion of their works to a vigorous attack upon the opinion said to be expressed in this extract, and severely criticise other cases containing similar expressions, they nowhere intimate that Justiee Park ever made such a ruling. In any event, and if either of the two species of evidence can be said to be in the abstract weaker than the other, the 'weight of opinion and the better reason is, that direct evidence is the more satisfactory.
As the number of witnesses, circumstances, etc., has no connec*345tion with the abstract value of the two species of evidence, it is as erroneous to rate circumstantial evidence in the ■ abstract higher than the direct testimony of one individual, as to rate it higher than the direct testimony of individuals. If it be contended that this instruction — or the opinion in People v. Cronin, 34 Cal., from which it was in part taken — merely affirms two propositions: first, that circumstantial evidence in the abstract is less satisfactory than direct testimony in the abstract; and second, that circumstantial evidence, as it usually occurs, is more satisfactory than direct testimony consisting of one witness, there are two answers. The instruction is not so guarded ; and if it were, the remark of Chief Justice Shaw in the Webster case, that it is not easy so to compare their relative value, applies. All that can be affirmed is, as he there said, “ that each has its advantages and disadvantages. The advantage in favor of direct is, that you have the direct testimony of a witness to the fact to be proved, who, if he speaks the truth, saw it done ; and the only question is whether he is entitled to belief. The disadvantage is, that the witness may be false and corrupt, and the case may not afford the means of detecting his falsehood.” On the other hand,i£ the advantages are that as the evidence commonly comes from several witnesses and different sources, a chain of circumstances is less likely to be falsely prepared and arranged, and falsehood and perjury are more likely to be detected and fail of their purpose. The disadvantages are, that the jury has not only to weigh the evidence of facts, but to draw just conclusions from them.” The result is that given from Starkie— not that the circumstantial evidence' is more satisfactory, but that it js so when the aggregate of doubt mentioned by him. is less than the doubt whether the direct witness be faithworthy. Further than this the law cannot go. It cannot even affirm that a jury is less likely to draw false inferences from proven facts than a single witness is to forswear himself, or mistake the evidence of his senses, in a prosecution for a public offense. If the instruction imputed to Justice Park is correct, is not also the converse ? Then, whenever a party is sought to be convicted on the direct testimony of one witness, must an acquittal be practically insured by instructing that such .testimony is “ much less satisfactory ” than would be *346circumstantial evidence of his guilt? We think that, standing alone, this concluding portion of the instruction is erroneous; nor do we see that the error is cured by what preceded. Throughout, circumstantial evidence is mentioned without ’reference to its strength or weakness as affected by the number of witnesses testifying, the number or • concurrence of circumstances testified to, or their nature as conclusive or inconclusive. Error being shown, the burden of establishing its immateriality is on the State, and in criminal cases, the showing of immateriality must be conclusive. We tíannot say that the instruction taken as whole might not have conveyed to the mind of a man of ordinary capacity an incorrect view of the law applicable to the cause. Such a juror, with this instruction in his hands, may have reasoned: “ It is true, I must not find the defendant guilty, unless the testimony has convinced me of his guilt- beyond a reasonable doubt; but this is a case of circumstantial evidence, and I am instructed that such testimony is more satisfactory than that of a single eye-witness would be. Now, if one of my neighbors — a man whom I have known and trusted for years — had sworn that he saw the defendant set fire to these cans, I should entertain no doubt of his guilt.” The very distinction made between the testimony of individuals and that of a single individual, increases this tendency to mislead, by inducing the construction that such testimony is often as satisfactory as the direct testimony of individuals, but it is always more satisfactory than that of one individual.
In Cicely v. State, 13 S. & M. 211, a case of circumstantial evidence, the Court below refused to instruct that, unless the jury were .as' well satisfied from the evidence of the guilt of the accused, as they would be from the testimony of a single witness testifying directly to the fact, they should acquit. The Court of Errors, sustaining the ruling, say : “ It is said (1 Starkie’s Ev.) that the legal test of the sufficiency of evidence to authorize a conviction is its sufficiency to satisfy the understanding and conscience of the jury : that a juror ought not to convict, unless the evidence excludes from his mind all reasonable doubt of the guilt of the accused.” This is doubtless the only true and practical criterion, by which the foree of evidence in criminal prosecutions, sufficient for *347conviction, is to be determined. The objection to this instruction is, that it applies a rule which is neither practical nor altogether safe.
Under the operation of this rule, the juror would be compelled to act not upon the direct effect which the evidence had produced upon his mind. He would be not only required to inquire into the state of his mental convictions; to ascertain whether the evidence, offered in support of the prosecution, had excluded from his mind all reasonable doubt; he would be forced to go farther, and to institute a comparison between the degree of conviction produced by the evidence and that which would be the result of the testimony of one direct witness; for that would be the standard by which he would have to determine the degree of certainty in the proof, which would authorize a conviction or require an acquittal. We have daily experience that the same evidence does not invariably produce the same degree of conviction in different minds. Hence, we may well conclude that the legitimate force of the direct evidence of a single witness would be differently estimated by persons whose minds were differently constituted. The practical application of the principle contained in the instruction would, in effect, be to adopt a standard fon estimating the force of this species of evidence, which would differ with the varying mental organization of each juror. Its practical effect, in all probability, would be, on the one hand, to lead to convictions in cases where, by the use of more intelligible and safe rule, acquittals would follow; and on the other, to produce acquittals where, by the same test, the parties would merit conviction. In the case cited, the jury had, as here, been already fully instructed on the question of reasonable doubt, and the charge refused was, and was admitted to be, as a legal proposition, an “ abstract verity.” The instruction given in this case is, as an abstract proposition, erroneous, but of a similar tendency to lead 'to improper convictions — to embarrass rather than assist the jury. Whether the testimony was sufficient in fact to authorize a conviction, is a question upon which we express no opinion. It is the right of the defendant to have that question decided by the verdict of a jury applying, and applying only, the proper legal, tests. Nor do we intimate, that in cases of this kind there is any impropriety in *348guarding the jury against the opposite errors of rating circumstantial evidence too low, and paying too much attention to arguments setting forth the danger of convicting innocent persons on such testimony ór giving instances of such convictions. To this end, it is proper that the jury be admonished in fitting terms, that it is es- ' sentially necessary to the security of mankind that juries should convict, when they can do' so safely and conscientiously, upon circumstantial testimony ; and that otherwise, the very secrecy with which crimes are often committed would secure safety to the criminal ; that in such cases the law allows the introduction of circumstantial evidence, and by it there may be, and often is afforded, a safe and satisfactory ground of assurance and belief; that the fact, that innocent men have sometimes been convicted on insufficient circumstantial evidence, does not prove that guilty men should not be convicted on sufficient circumstantial evidence ; any more than the fact, that innocent men have been convicted on the"direct testimony of perjured or mistaken witnesses, proves that guilty men should not be convicted on adequate direct testimony free from’ suspicion of bias or mistake ; that the force and efficiency of either species of evidence may, according to circumstances,-be carried to an indefinite and unlimited extent; that circumstantial évidence, as well as direct, may be so cogent as to exclude from the mind all reasonable doubt of the guilt of the accused, and that if the jury is convinced by the evidence of such guilt, beyond such doubt, it- is equally their duty to convict, whether the result is produced by direct or by circumstantial evidence. The advantages and disadvantages of each mode may be pointed out, and the jury instructed that the just conclusion to be drawn from cases where innocent persons have been convicted upon circumstantial testimony, is not that we are to abandon all confidence in this kind of evidence, but that such cases may be addressed as an admonition to the jury to use great care and caution in weighing the evidence, and to draw no inferences from the proved facts but such as are reasonably and morally certain.
We are asked to review the evidence, and to decide that it does not sustain the verdict. ' But on the threshold of this inquiry we are met by the question whether this is. within our jurisdiction. *349This question has not been argued, and we do not propose to decide it in this ease. We call attention to it because of its importance, and that when it next arises it may be fully argued. To hold that this is a question of fact may amount to a denial of our jurisdiction in all cases where the granting or refusing a new trial depends upon conflicting or contradictory affidavits, as in case of alleged misconduct of jurors: such cases, for instance, as People v. Plummer, 9 Cal.
By the Constitution, our jurisdiction in’criminal cases is limited to questions of law alone. The statute in force before, at the time of, and since the adoption of the Constitution — every lineament of which shows its common law parentage — confines our jurisdiction within the same limits. We must then look to the common law to ascertain what are questions of law and what questions o| fact. By the’ common law the Judges answered the former, the jury the latter.. The fact must, it is said, in the nature of things,' be first ascertained — “ without a fact agreed it is as impossible for a" Judge, or any other, to know the law relating to .that fact, or direct concerning it, as to know an accident that hath no subject.” When the fact is disputed, the jury must find it. Ordinarily, in so doing, they compound their verdict of the law as given to theni by the Judge, and the fact as ascertained by themselves. But there are methods by which the party may, if he choose, withdraw from the jury the application of the law to the fact. This may be done by a demurrer to' evidence ; or he may ask an instruction that there, is no evidence before them of a certain essential ;fact, or no evidence to sustain the issue on behalf of the other party, and tender an exception if the instruction be refused. By either method a question of law is raised, but all the facts which the evidence tends to prove are admitted. Then, by parity of reasoning, are we not to look at the evidence on this motion just as we would on a demurrer to evidence, or a bill of exceptions to a refusal to instruct as to the insufficiency of the whole evidence ? If so, is every fact and conclusion* which the evidence in the slightest degree tends to prove admitted, or only such as it fairly and reasonably tends to prove ? It is said there is a “ sufficiency in law” as well as a “ sufficiency in fact ” of evidence. Its suffi*350ciency in fact being exclusively for the jury to decide, subject only to revision, as a “ question of fact,” on. motion for a new trial; not on a writ of error, as a “ question of law not the subject of a bill of exceptions, which raises only questions of law decided during the trial. But that the legal sufficiency of the evidence adduced to sustain the issue or to establish any essential fact, is a' question of law and not of fact; and wherever it is so light and inconclusive that no rational, well constructed mind can infer from it the fact which it is offered to establish, it is the duty of the Court, according to circumstances, either to reject it as inadmissible or to instruct the jury that there is no evidence before them to warrant their finding the fact thus attempted to be proved. According to this view, we have jurisdiction — all the evidence being embodied in the bill of exceptions or affidavits — assuming all facts of which there is legal evidence and drawing all conclusions which can be fairjy and logically drawn in support of the verdict or decision — to decide, as a question of law, whether the evidence is sufficient to sustain such verdict or decision in a criminal case.
It is frequently said that the granting or refusing a new trial, on the ground that the verdict is against the weight of, or contrary to, the evidence, is a matter of discretion; and that therefore such rulings cannot be reversed on a writ of error, or spread on the record by a bill of exceptions. Perhaps this is only another way of saying that it is the decision of a question of fact — the determination of a question of law hardly admitting of the exercise of discretion.
If our jurisdiction is more limited in criminal than in civil cases; does not this impose on District Judges the obligation of a greater care and caution in thus finally deciding on questions involving life and liberty; of attending more strictly to the distinction between an appellate Court, which only sees the evidence as embodied in a transcript, and a nisiprius Judge who, equally with the jury, closely watches the testimony as it is given in, observes the bearing and demeanor of witnesses, and in short, possesses the same facilities for arriving at a correct ascertainment of the facts as the jury ? While the former tribunal can only — even in a civil case — grant a new trial where the record can show that the verdict is palpably against the weight of evidence, the latter should, wherever there are strong *351probable grounds to suppose that the merits have not been fairly and freely discussed, and that the decision is not agreeable to the justice and truth of the case, though they should not, where the scales of evidence hang nearly equal: that which leans against the verdict ought always strongly to preponderate. That is, in a civil case, the nisi prius Judge should not grant a new trial, where, it is doubtful whether the verdict is right or wrong, where his mind wavers and oscillates; but' should, where he has a clear and unhesitating conviction that the verdict ought to have been the other way, and in arriving at this conviction he can consider matters not appearing on the record, such as his observation of the bearing and demeanor and mode of testifying of the witnesses, and other such circumstances as cannot be preserved in the record. It is essential to the beneficial working of jury trials that this power should be fearlessly exercised, even in civil cases ; in criminal cases the. only distinction would seem to be, that the Judge need only be satisfied that the evidence does not establish guilt beyond a reasonable doubt. It is worthy of consideration too, whether, in deciding all motions in criminal cases, the District Judges should not adopt the practice of stating which witnesses or affidavits they believe, in whole or in part, or. separately their findings of fact and conclusions of law, so as to secure to the defendant his constitutional privilege of an appeal on questions of law. Such a practice, in Louisiana, seems to have followed the construction put upon language in their Constitution, exactly the same as that used in ours, their Courts having decided that whether the Judge below credited an affidavit, or whether the defendant used due diligence in procuring the attendance of witnesses, &c., were questions of fact, and could not be reviewed on appeal, any more than the refusing a new trial for verdicts against evidence. The following authorities bear more or less on the points suggested, but not decided in this opinion: Whiton v. Nichols, 3 Allen, 586; 8 Cushing, (Mass.) 74; Comm. v. Merrill, 14 Gray, 416; State v. Nelson, 3 La. Ann. 500; 2 Ib. 837; 16 Ib. 308; 14 Ib. 673; House v. Elliott, 6 Ohio State, 499; 7 Ib. 299; Stephens v. White, 2 Wash. (Va.) 203; Bulkeley v. Butler, 9 E. C. L. Rep. 198; Vines v. Corporation, 1 Younge & J. 6; Coffin v. City, 26 Iowa, 520; 22 Ib. 241; 15 Ib. 90; Clarke *352v. Dederick, 81 Md. 150; 7 G & J. Md. 20; Avery v. Bowden, 88 E. C. L. R. 972; People v. Lewis, 36 Cal. 533; People v. Jones, 31 Cal.; People v. Renfrew, Ib. Jan. 1871; 10 Ib. 301; Ib. 195; 27 Ib. 500; Denny v. Williams, 5 Allen, 1; Gibson v. Hunter, 2 H. Blackstone, 205; 23 N. Y. 344; 19 Ib. 212; Parsons v. Bedford, 3 Peters, 448; 12 Ib. 345; 2 Ib. 623; 12 Wheaton, 180; 5 Wend. 114; 18 Ib. 84; People v. Adwards, 5 Mich. 22; Jacob v. U. S., 1 Brockb. 527; 1 Rawle, 432; Hansbrough v. Thorn, 3 Leigh, 147; Graham v. Camman, 2 Caine’s R. 168; Winsor v. Queen, Law Rep. 1 R. B. 289.
We cannot; on this record, say that, as a question of law, the testimony was insufficient to justify a conviction. The judgment and the order refusing a new trial are reversed, and the cause reminded for a new trial.