Respondent was convicted in Ionia county of the murder of his wife by poison, on May 9, 1882. The information sets out that the death was caused by a series of acts of administering arsenic with her medicine, beginning April 25, 1882, and continuing till her death. The body was taken up by the coroner on the 22d day of August thereafter, when a partial examination was had by local physicians, and the stomach and part of the rectum were removed and placed in one jar, and part of the liver and of one kidney placed in another jar, and these jars sent to Dr. Prescott, at the University, to be analyzed. He made an analysis in two different parcels of portions of the contents of each jar, making a single analysis of part of the stomach and rectum, and a separate one of part of the liver and kidney, but not taking any one organ by itself. On a subsequent occasion he made another examination of the brain, and of parts of the muscles of the calf of each leg. He found no trace of arsenic in the brain or calf. In each of the other analyses he found considerable quantities, estimated at from six to fifteen grains in the entire liver and one kidney, and fifteen grains in the stomach and rectum.
Upon the trial the testimony whereby it was sought to prove the death by arsenic, and the responsibility of respondent for causing it, was all circumstantial. It involved inquiries into the question whether the arsenic discovered was the real cause of the death as introduced into the body during life, and feloniously, and into the symptoms of the illness of deceased as those of disease or of poisoning by arsenic. Beyond this it covered testimony of the conduct of respondent, and some relations allowed to be gone into on the question of motive.
Upon the argument in this Court we called upon the counsel for the People to sustain the rulings complained of by the assignments of error, and being satisfied that the *67rulings were not in conformity with previous decisions of this Court, we dispensed with any full oral argument for the respondent, beyond his brief.
¥e do not propose to consider the minor questions of error, so far as they are not likely to involve difficulties hereafter. The jury lists will not probably be defective in the future, or defective in the same alleged way. Neither is it likely that further witnesses will be admitted hereafter of whose names and residences respondent has had no sufficient notice. And most questions of the order of proof can be disposed of in a general reference, without taking up all of the assignments singly.
In the case of People v. Hall 48 Mich. 482, we had occasion to remark upon the necessity, in homicide cases particularly, of keeping closely to the established rules, and of allowing no evidence of motive, and no other testimony beyond the main issue of the crime itself, until the testimony concerning the corpus delicti- has been completed. In the present case, while there was some departure from this order in other respects, the most important was in allowing testimony concerning illicit relations between respondent and a woman who had once lived in the neighborhood at a period considerably earlier than the wife’s sickness, to be shown, not only before the other testimony was all in, but without any proof of such relations at about the time of the alleged crime. This was done, it is true, on the promise that the relations should be shown to have been continued-. This however was not shown, and the subsequent testimony allowed to come in on this subject not only had no tendency to prove it, but rather the reverse. All the legitimate evidence concerning respondent’s domestic relations indicated entire harmony between husband and wife. The court refused to strike out this improper testimony, and laid stress upon it to some degree in the charge.
Such testimony, if received, has a manifest tendency, as it must have been designed, to produce a disposition in the jury to lean against the respondent, and to put an unfavorable construction upon his acts. As suggested in People v. *68Hall, there is no possible reason for allowing such testimony to be put in out of its order. The prosecution, under our system of criminal practice, is bound to proceed against no one without having in advance the testimony on which a conviction is to be sought; and while in a civil case it may be sometimes allowable to receive testimony on a pledge that it will be made admissible by future evidence, it never should be done, even there, if it is likely to be seriously prejudicial. A jury cannot readily dismiss from the mind, so that it will leave no trace, testimony that blackens character or impugns motives. In a criminal case the prosecution should be allowed no such latitude. The mischief is effected when the testimony comes in, and gives color to all the subsequent views of the jury, so that they look suspiciously upon all that is said or done by the respondent which is open to any consideration of motive.
It was, we think, improper to exclude evidence of the transactions in the house between the death and burial. They were important in the inquiry concerning the use of antiseptics, which was then known to be material. A similar principle applies to the' dealings and transactions concerning the employment and understanding as to Drs. Nay and Epply. It was error to exclude the evidence as to the contents of the page of the book shown by Mrs. Wortman to respondent. It was the same as if she had read or spoken the same words. It was also error to shut out the impeaching evidence concerning Mrs. Wortman’s attempt and failure-to get employed in respondent’s house after the death of his wife. It bore on her motives, and to some extent was contradictory of her testimony as believing in his guilt. Geary v. People 22 Mich. 220; Hamilton v. People 29 Mich. 173.
But the more important questions, inasmuch as they are inseparable from the main issues in the case, arise upon the course of testimony depending on the opinions of scientific-witnesses,- and upon the burden of proof, on which we think there was fatal erroz\ Several of the difficulties presented on the record must have arisen from a failure to appreciate-the decision in the case of People v. Hall, and other cases *69involving similar inquiries. These matters can be more satisfactorily dealt with together, so far as they resemble each other in principle. To explain them it will be necessary to recall the precise character of the controversy.
The analysis having revealed the presence of arsenic in such quantities that if administered during life it would have been sufficient to produce death, and the fact that it was so found not being of itself indicative of when or how it was introduced into the substances delivered to Dr. Prescott to be analyzed, there were two lines of inquiry connected with this particular question. One was whether the position and proportions of the arsenic found, indicated satisfactorily absorption during life through the organs which would serve as vehicles for that purpose; and the other, whether it could have been introduced into the body after death, and reached the various organs where it Avas found, by any means that Avonld be practicable under the circumstances. In this connection we may remark that the undertaker, who was alloAved to give an opinion on a physiological question connected with this subject, presented no claims entitling him to give an opinion as a scientific expert, and his testimony was improper so far as it related to anything but specific facts. A third possibility was not argued, and it is not suggested that the facts call for its application; that is, the careless or intentional introduction of poison into the material for analysis. "While the testimony does not indicate the utmost care in the removal and packing of the contents of the jars at the time when the body was exhumed, the record gives no reason for supposing that there was any lack of precaution that involved danger of improper intermeddling. The controversy, therefore, included only the other íavo inquiries.
But beyond this there was a medical question upon which the case very largely depended, and that was whether the symptoms of the fatal sickness Avere certain indications of poisoning and its effects on the system of the deceased. If such was beyond doubt the character of the symptoms, the ambiguity of the other post mortem appearances and discoveries would be greatly lessened or removed. If the S}’mp-*70toms were ambiguous, or if they belonged to some disease or malady, they might raise such probabilities as to negative guilt, or such reasonable doubts as would make it improper to convict.
Upon this last question of the character of symptoms, an important class of assignments of error is presented, which substantially relate to the burden of proof. While properly laying down in various places the general doctrine requiring the jury to convict only where they had no reasonable doubt, the effect, nevertheless, of several rulings was to indicate— intentionally or unintentionally — that testimony introduced by the defendants upon this particular class of appearances and their meaning, was in the nature of an affirmative defense, and that the People were not bound to exhaust their proof so as to negative all results but poison. This led to the allowance of more or less testimony as rebutting-, which respondent claimed belonged to the principal case, and to the natural inference (and comparing the charge and refusals to charge with some other rulings on testimony, to the necessary inference) that the defense had the burden of proof to meet the presumptions supposed to be raised by the prosecution.
This idea is not correct. In every criminal case the burden is throughout upon the prosecution. Whatever course the defense deem it prudent to take in order to explain suspicious facts or remove doubts, yet it is incumbent on the prosecution to show under all circumstances, as a part of their own case, unless admitted or shown by the defense, that there is no innocent theory possible which will, without violation of reason, accord with the facts. And in a case of alleged poisoning, where the symptoms and appearances during the last illness become controlling facts in determining whether the death was from poison or from disease, the charge is not made out unless the prosecution negative everything but poison as the cause of death. And this they can only do by showing affirmatively that the combined symptoms, and the absolutely certain facts with which they are associated, are inconsistent with any other disease or ailment-*71It has been generally regarded as unsafe to convict on symptoms alone, from the danger of confounding the appearances of disease and poison. But it is little if any less dangerous to allow the presence of poison in a dead body, where there is no certainty whether it got there before or after death, to be given a presumptively guilty significance by symptoms which are themselves open to an innocent meaning. We doubt very much whether the learned judge who tried the case intended to hold differently, but we think the record indicates that such was the effect of his rulings.
The remaining questions of much importance arise out of the scientific testimony, and the facts and theories to which it was applied. To understand the bearing of this branch of the controversy, it becomes necessary to notice the principal features of the scientific discussion. The differences were radical on some important issues, and arose in part, beyond question, from the absence of some means of judgment which it was desirable and in some respects necessary to furnish in order to reach a safe conclusion, and in part from inferences of witnesses from facts not within the hypothetical statements.
The symptoms which were claimed as . indicating poison were shown to be marked and definite as early as between the 18th and the 25th of April, and to have continued day after day until the 9th of May, during which day the death occurred in the afternoon. The respondent is claimed by the prosecution to have administered or caused to be administered daily, and several times a day, as early as April 25, and thereafter, doses which it is insisted were poisonous, in amounts, which if they were arsenic in any considerable proportion, were each large enough, according to the medical testimony of how much will kill, to be a fatal dose. The amount found in the organs examined was larger than the testimony of some, if not of all, of the experts shows is absorbed during life in most cases of arsenical poisoning hitherto known and retained in the various organs, and according to some of the testimony more than it is understood could possibly be so absorbed and retained. The time within which doses large enough to be *72fatal will kill is shown by much of the testimony to be much shorter than some of the facts in this case would indicate to have elapsed, if this was a case of poisoning. The expert testimony is equally positive, on the part of some of the strictly scientific witnesses, against the possibility of harmonizing the discoveries and appearances with slow or chronic poisoning, or with any poisoning.
In regard to the meaning of the symptoms themselves, if the witnesses all understood the same interrogatories alike, there was a great difference of opinion. It is claimed by the defense that much of this conflict and difficulty is due to the course of examination on the trial, and upon this they are probably right. There was, at any rate, a departure from the rules heretofore laid down, both in the questions allowed to be put and in the method of answering permitted. And thisvdiffieulty was increased by permitting inquiry where the foundation for correct inquiry was lacking.
It appeared from the testimony that, during the illness of deceased, besides the temporary visit of one physician, she was attended by two other physicians, who seem to have differed somewhat as to the precise measure of their responsibility, but who had the means of knowing her symptoms and the course of the disorder, and who at no time took any stops to deal with it as involving poison. Upon the trial it was intimated that one or both had suspicions that their treatment was counteracted, and that if so, it was by arsenic. Neither of them took any measures to verify the suspicion, if entertained, or to give antidotes. No one, either before or after death, is shown to have examined the vomit or other discharges, or to have taken any other measures of caution or detection. After the death, one of them, when suspicion had arisen among others, discouraged having a post mortem examination as unnecessary, and neither took measures to procure one. It is clear that such measures as would naturally have been suggested to a physician by a suspicion of the kind referred to, would have settled the nature of the disorder beyond doubt. The duty of a physician to save life, and the duty of a citizen to prevent and expose crime, are so mani*73fest that we are bound to suppose the suspicion was not entertained definitely, or that the symptoms could mot have appeared plainly indicative of poison at the time. At any rate the lack of securing the obvioús means of detection is due to witnesses who on the trial testified as experts upon some points which were very material, when their conduct during the sickness betrayed no suspicion.
The same difficulty as to various indications to be sought on a post mortem inquiry arose in making that inquiry. It was then known that respondent did not dispute the probability that poison would be found, but referred it to the fact that poisonous embalming or preserving fluid had been injected through two channels into the body. The importance of then examining the various organs and parts of the body to look and see whether such appearances existed as would have been produced by the disease which it had been assumed was the cause of death, was obvious to any mind. These symptoms began before any charge is made that poison was given, and so early as to raise serious questions whether, if caused by poison, death could have been so long delayed. But no thorough search seems to have been made, and attention was chiefly, although not entirely, given to those parts which it was deemed important to analyze for poison. And although the brain was afterwards analyzed, it does not appear to havebeen carefully inspected for other purposes. No examination was made of the lungs to ascertain whether arsenic had become deposited there, as upon one theory of the experts was possible, if not probable; and other parts, concerning the significance of which there was no difference of opinion, were passed over slightly. And in separating the organs for examination by the chemist they were so assorted as to render it. absolutely impossible, as the analyst testified, to tell how much arsenic was in either, or whether all discovered in the mixture of two parts might not have been in either alone.
This absolutely destroyed the value of' the analysis for the purpose of indicating by itself whether or no the poison was introduced and absorbed in life. The testimony indicates that arsenic deposited in the various organs, when taken dur*74ing life, is not, when found after death, found in similar quantities or proportions in all, and that the proportionate character of the deposits is a very important factor in scientific conclusions. Some of the significant organs were left out entirely from any analysis, and the others so confounded that no safe inference could be drawn from them. And this defect rendered more or less of the testimony to rest on such pure guess-work as to render it improper for drawing some conclusions that were allowed to be drawn, and left as the only fact ascertained, which was important at all, the fact that a certain quantity of arsenic was found, which, so far as chemical science could detect, as the chemist testified, might have been put in the body either before or after death, and had, therefore, no legal significance, standing alone. But, as indicating a quantity which, according to a considerable part of the scientific testimony, was in excess of what had ever been found in certain of the organs in cases of death from poison, and as further indicating, from the same testimony, the improbability that any guilty poisoning could have dated back so far as charged, it bore very strongly under those proofs against the theory that the symptoms which continued so long alike were poison symptoms.
We are obliged, therefore, to see that the case actually depended almost entirely, if not absolutely, upon the conclusive character of the symptoms; and as the only proof that respondent committed any wrong was based on the inference that he substituted poison for other medicine, without any direct proof on that subject, the experts become the witnesses on whose testimony the whole case rests. Any error in allowing expert opinions was therefore a fatal error.
Of this testimony there were two branches, namely, the chemical and the medical. An expert in one of these is not necessarily an expert in the other. In the present case the four principal experts appeared in both capacities. The line does not seem to have been drawn as closely as it should have been in examining the experts generally.
Objection was made to several questions as asking opinions on defective or improper assumptions. As this same sub*75jeet was discussed in People v. Hall to some extent, it is not necessary to dwell upon it at large. It is sufficient to say that in some questions assumptions of fact were made, supported by no evidence ; the actual administration of poison was assumed, when it was only deducible by inference; symptoms were detailed without any order of time, or any such connection of antecedent and subsequent occurrences as could give them any significance which would be definite; and no such certainty was called for as would enable the jury to know on what basis of actual evidence on the facts shown by the witnesses who were not experts, the experts themselves were testifying. It also appeared on several occasions that the experts had assumed facts not submitted to them for their opinion, but culled by themselves from what they had heard during the trial, and with which they had no right to meddle for themselves.
It is hardly necessary to say that such testimony is not competent. Another source of error quite as mischievous, and which the court evidently desired to prevent, but did not always do so, was the introduction of what is no more than hearsay evidence, in the shape of references to writers and books in such a way as to invoke their authority. As we have had occasion on more than one record to explain heretofore, expert evidence is only admissible on the theory that the jury cannot be supposed to comprehend the significance of facts shown by other testimony which needs scientific or peculiar explanation by those who do comprehend it. But this does not permit hearsay testimony of the written or spoken opinions of other persons, whom the jury have no means of examining as to their learning, their honesty, or their sources of special knowlege. Every medical and scientific writer bases much of his conclusions upon what he believes to be true in the reported facts and opinions of other men of science. Those facts may be correctly stated, or they may be assumed on small or no foundation. Those opinions may be taken carelessly at second hand, or they may have been thoroughly weighed before adoption. No one can tell whether a medical book or opinion is reliable or *76not, until lie has applied himself with some fitting preparation to its study and criticism. The book may be good in part and bad in part, and neither court nor jury can presumptively ascertain its quality.
No one has any title to respect as an expert, or has any right to give an opinion upon the stand, unless as his own opinion; and if he has not given the subject involved such careful and discriminating study as has resulted in the formation of a definite opinion, he has no business to give it. Such an opinion can only be safely formed or expressed by persons who have made the scientific questions involved matters of definite and intelligent study, and who have by such application made up their own minds. In doing so, it is their business to resort to such aids of reading and study as they have reason to believe contain the information they need. This will naturally include the literature of the subject. But if they have only taken trouble enough to find or suppose they find that certain authors say certain things, without further satisfying themselves how reliable such statements are, their own opinions must be of very moderate value, and whether correct or incorrect, cannot be fortified before a jury by statements of what those authors hold on the subject. The jury are only concerned to know what the witness thinks, and what capacity and judgment he shows to make his opinion worthy of respect. If the opinion of an author could be received at all, it should be from his own words — not in single passages, but in combination; and this, as has been heretofore held, cannot be done. It is excluded chiefly as both unknown as to value and as hearsay, and an attempt to swear to his doctrine orally would be hearsay still further removed, besides involving the other difficulty of needing interpretation and responsibility.
A large part of this scientific evidence consists of repeated references for support and confirmation to the statements of authors, and as might be expected, the different witnesses have not always read through the same glasses. To settle their differences the only resort must have been to the very books which no one claims were admissible. These refer*77enees were not merely made on cross-examination to test the knowledge and veracity of the witnesses, but came in as frequently on direct examination. In Pinney v. Cahill 48 Mich. 584, a witness who had asserted that a certain book laid down a certain proposition was allowed to be contradicted by showing it did not. But it is questionable whether the original assertion was properly drawn out, although when made it was proper to let it be contradicted. It was distinctly held in Marshall v. Brown 50 Mich. 148, that attempts to evade the excluding rule by examining or cross-examining in such a way as to get an opportunity to get books before the jury could not be permitted.
Upon medical questions, while persons may testify whose knowledge is chiefly theoretical, there can be no question of the superior value of practical knowledge, combined with theoretical, especially on such matters as involve the interpretation of symptoms and actions of the sick. It is also a rale which has been laid down as one of law, that no amount of theory can overcome facts established by credible testimony, and that juries have no right to form their own opinions upon matters not proven or laid before them, in disregard of the testimony. Treat v. Bates 27 Mich. 390; Wood v. Barker 49 Mich. 295.
We have thus indicated our views upon the matters which, upon the argument, we deemed most important. There are other matters which are not so important, upon which we do not deem it necessary to pass. We cannot however fail to notice that, in addition to many grossly leading questions, in the excitement of the trial things were said before the jury and to witnesses which were not proper, and which it was not good practice to tolerate. The record also contains a great deal of prolixity and irrelevancy which is productive of expense, wrong and oppression to parties compelled to defend themselves against criminal accusations and to remove judgments for review. Where all witnesses are known, and most are expected to have been sworn on the examination, there can be no occasion for prolixity in examining them on the trial.
*78.The judgment must be reversed, and the prisoner remanded to Ionia county for such further action as the authorities desire to take in the premises.1
The other Justices concurred.Upon a second trial in December 1884, the respondent was acquitted.