Ladd, J.
It is enough now to say that the evidence was such as to leave the question of defendant’s guilt of be.getting complainant’s child, then 9 months old, at the least •doubtful. Her story implicating him, appears not very probable, and her admission of having had intercourse with Waller a year previous to its conception, and sleeping alone in •a room accessible to. the latter continually up to that time, indicates, notwithstanding her denial of repetition, the pos■sibilily of its having been his offspring. So that the intro■duction of the immature child in evidence “for the jury to look at; * * * to examine as to the identity and re■semblance between the baby and putative father” — as stated by counsel for the state — may well have played an important part in settling the controversy. The color of its eyes and hair, its complexion, the contour of the brows .and shape of hands, any or all of which may have related back three or four generations, doubtless were given weight in making this •comparison. What they were we have no means of know*417ing, nor does this record disclose in what respects these differed from or resembled the hair, eyes, complexion, brows, or hands of defendant, or of Waller. Thus, the jury based their verdict in part at least, on their individual knowledge of facts, or opinions resting on facts, of which this court, on appeal, can acquire no information, making of themselves “silent witnesses in the case, burdened with testimony unknown to both parties, and in respect to which no ■opportunity for cross-examination or correction of errors, if any, could be afforded either party.” See Close v. Samm, 27 Iowa, 507; Washburn v. Railway Co., 59 Wis. 370 (18 N. W. Rep. 328). In such a case might a new trial ever be ordered because of the insufficiency of evidence ? Nevertheless this court in State v. Smith, 54 Iowa, 104, held that ■a child 2 years and 1 month old might be exhibited to the jury, though to exhibit o>.e of 3 months had been adjudged an error in State v. Danforth, 48 Iowa, 43. While conceding that “resemblances often exist between persons who are not related, and are wanting between persons who are,” the ruling seems to rest on the proposition that “what are called 'family resemblance’ are sometimes so marked as scarcely to admit of a mistake.” With respect to proper age, it wa& said that “a child which is only 3 months old has that peculiar immaturity of features which characterizes an infant ■during the time that it is called a 'babe.’ A child 2 years old •or more has, to a large extent, put off that peculiar immaturity.” If this is to be the criterion, then, surely, a child •of 9 months is toó immature to afford aid to the jury in settling its paternity. True, resemblances then are frequently imagined. But what one will construe as a similarity, another, with the same knowledge of the parties between whom the comparison is made, will be unable to detect. If alike in some respects, they differ in others. It is all a matter of notion, fancy, or guesswork, and ought to be given the slightest weight in determining an issue fraught with such *418grave consequences. In People v. Carney, 29 Hun, 47, the court, observing that children of the same family have eyes and hair of different colors, declared that it is “a dangerous doctrine to permit a child's paternity to be questioned or proven by the comparison of the color of its hair or eyes .with that of the alleged parent.” In Hanawalt v. State, 64 Wis. 84 (24 N. W. Rep. 489), the exhibition of a child under a year old was held to have been improper, the court saying: “In any case this kind of evidence is inherently unsatisfactory, as it is a matter of general knowledge that different persons, with equal opportunities of observation, will arrive at different conclusions, even in the case of mature persons, where a family likeness will be fully developed if there be any; and, when.applied to the immature child, its worthlessness as evidence to establish the fact of parentage is greatly enhanced, and is of too vague, uncertain, and fanciful a nature to be submitted to the consideration of a jury.” As opposed to such exhibitions, see, also, Clark v. Bradstreel, 80 Me. 466 (15 Atl. Rep. 56, 6 Am. St. Rep. 221); Risk v. State, 19 Ind. 152; Reitz v. State, 33 Ind. 187; Ingram v. State, 24 Neb. 33 (37 N. W. Rep. 943). ^Vithout expressing an opinion as to the correctness of State v. Smith, supra, this court is not prepared to extend the rule there approved, and sanction the exhibition of a child under 2 years of age to the jury, as affording any aid in ascertaining its parentage; and in this class of cases, where, as is well known, the feelings and sentiment so often enter into the contest, the exhibition of the fruit of the unlawful relation cannot have been otherwise than extremely prejudicial. An exception may, however, exist where the parents are alleged to be of different races. Garvin v. State, 52 Miss. 207; Warlick v. White, 76 N. C. 176. In Gilmanton v. Ham, 38 N. H., 108, and Finnegan v. Dugan, 14 Allen, 197, the age of the child exhibited does not appear, while comparison at any age was upheld in Gaunt v. State, 50 N. J. Law, 490 (14 Atl. Rep. 600), and State v. Woodruff, 67 N. C. 89. See, *419also, Jones v. Jones, 45 Md. 151. These decisions seem to have been influenced somewhat by the ruling of Lord 34 airfield in the Douglass Case, before the house of lords in 1169, allowing resemblance of adults to be shown. That marks of family resemblance often exist between adults and even mature children may readily be conceded, but it does not follow that this is generally true of nursing or immature babes, with unsettled features, and peculiar characteristics undeveloped. — Reversed.