MR. JUSTICE MYERS

delivered the opinion of the court.

This is an action for damages. In the court below, plaintiff had judgment, on verdict for $20,000, against the defendant company. It made a motion, based on various grounds, for a new trial. Included in the grounds were “misconduct of the jury” and “excessive damages.”

*146 At the hearing of the motion, only those two grounds were urged. As to the first mentioned, it was contended the misconduct of the jury was that of arriving at the verdict by resort to the determination of chance and returning a quotient verdict. In support of that contention, the moving defendant filed the affidavits of five of the jurors. In opposition, plaintiff filed the affidavits of six other jurors. The trial court granted the motion for a new trial and did so expressly upon the ground of the alleged misconduct of the jury. Plaintiff appealed.

At the hearing in this court, there were urged by defendant company, in support of the action of the trial court, only the two grounds urged at the hearing in the trial court. We take up, for consideration, that of the alleged misconduct of the jury.

At the outset, counsel for plaintiff urged that, the hearing upon this ground, in the trial court, having been had wholly upon affidavits, we should give no weight to and should indulge in no presumption in favor of the order of the trial court in granting a new trial. Counsel contend we are in as good a position to analyze and pass upon the affidavits as was the trial court. It may be so and we address ourselves to the task. We admit the same reasons do not exist why there should be a strong presumption in favor of the trial court’s ruling as exist in cases in which facts are determined upon conflicting oral testimony but we cannot concede the ruling of the trial court, in this matter, should be ignored, in our consideration. In two leading Montana cases of this character, decided wholly upon affidavits, as to the question here at issue, this court spoke approvingly of the discretion of the trial court in determining, by its order, conflict between contradictory affidavits. “That contradiction was resolved by the district court in favor of the McNamara affidavit.” (Gordon v. Trevarthan, 13 Mont. 387, 40 Am. St. Rep. 452, 34 Pac. 185.) “We enter upon our consideration of this appeal indulging in the presumption that the trial *147 court’s ruling was correct and the burden is upon the appellant to present a record which overthrows that presumption.” (Great Northern Ry. Co. v. Benjamin, 51 Mont. 167, 149 Pac. 968.)

A great many decisions are to be found upon each side of the question of what is a quotient verdict and the quantum of proof required to establish that a verdict is one of that character. There is not much difference about the rule of law as to what is a quotient verdict. Generally each case is decided upon the facts of the particular case, as to whether or not it comes within the rule.

In Great Northern Ry. Co. v. Benjamin, supra, this court adopted this rule, taken from 29 Cyc. 812: “A verdict will not be set aside merely because the amount thereof was the result of a compromise between jurors nor because th,e amount was first found by adding together the amounts the several jurors thought should be given and dividing the sum by twelve, if there was no agreement in advance to return a verdict for the quotient so found. Where the jurors agree in advance to be bound by a quotient so determined or where the consent of any juror to a verdict is determined by any resort to chance, a new trial must be granted.” The earlier Montana case of Gordon v. Trevarthan, supra, adopted from Thompson & Merriam on Juries, par. 408, et seq., this rule: “The facts vitiating such verdicts are the agreement by the jurors to go into the process of marking amounts, adding them and dividing the same by 12 and the agreement that the result so obtained shall be the verdict, without further consideration; and the fact that such proceedings were taken by the jury in pursuance of such an agreement and that the result so obtained was returned as the verdict.” Those pronouncements conform quite fairly to the general rule recognized throughout the land.

Our statute, giving grounds for a new trial, section 9397, Bevised Codes, 1921, in subdivision 2, says: “Misconduct of the jury; and whenever one or more of the jurors have been *148 induced to assent to any general or special verdict, or to a finding on any question submitted to them by the court, by a resort to the determination of chance, such misconduct may be proved by the affidavit of any one of the jurors.” It is now established beyond controversy that what is known as a quotient verdict is determination of chance and is misconduct of the jury.

From the wording of our statute, it appears it is not necessary for all of the jurors joining in a verdict, nor for a majority of them, to be induced by chance method to assent to a verdict. It appears that if any one of the jurors joining in a verdict be so induced to assent it is misconduct of the jury and this court has so held. The opinion in Great Northern Ry. Co. v. Benjamin, supra, says: “Where the consent of any juror to a verdict is determined by any resort to chance, a new trial must be granted.” The opinion in the earlier Montana case of Gordon v. Trevarthan, supra, says: “It appears from McNamara’s affidavit that at least one juror- — that is, himself — was induced to assent to this verdict by reason of the quotient proceeding; so it would seem that this, under the statute, is enough to vitiate the verdict.” Other jurisdictions, having the same statute, hold likewise. (City of Ottawa v. Gilliland, 63 Kan. 165, 88 Am. St. Rep. 232, 65 Pac. 252; Galveston, H. & S. A. Ry. Co. v. Brassell (Tex. Civ. App.), 173 S. W. 522; Wright v. Union Pac. R. Co., 22 Utah, 338, 62 Pac. 317; Williams v. State, 15 Lea (Tenn.), 129, 54 Am. Rep. 404.) The citations could be greatly extended. It is the general rule. (29 Cyc. 813.)

Application of the foregoing stated rule would simplify consideration of the ease at bar. In this case, nine jurors returned and assented to the verdict. To be valid, it is required that at least eight must have returned and assented to it. Not applying the stark rule that, if one of the nine assenting jurors was induced by resort to the determination of chance to assent to the verdict, it is void, in this case we shall be more liberal and say, if it may be shown that two *149 of the nine assenting jurors were so induced to assent, that Would reduce the number freely and properly assenting to seven, one less than is required by statute to return a legal verdict, in a civil action, in district court; and that undoubtedly would vitiate the verdict and render it not a lawful verdict — not lawful because assented to by one or more jurors by resort to determination of chance, and not lawful because not freely and properly returned by eight jurors, the number required by law.

Turning to the record, we find the five affidavits of jurors filed by defendant and the six, filed by plaintiff. An analysis of the affidavits shows some differences and some confusion among the affiants, in respect to the matter under consideration, as to just what occurred in the jury-room. That is common. In many of these “quotient-verdict” cases, there are differences of recollection among jurors, shown by their affidavits or oral testimony, as to just what did occur.

In the record before us, we find that nine of the affiants declare there were had two quotient processes of writing sums, adding and dividing. Some of them give the result of each; the first, they say, resulting in an amount something over $14,000 and the second, something over $19,000. Others of those nine affiants mention the result only of the second computation, something over $19,000, and, while referring to the first computation, do not give the result in figures. Two of the affiants make mention of only one computation, that which resulted in an amount something over $19,000. They do not deny there was another computation but simply do not mention it. Whether they had no recollection of it or simply omitted to narrate it we do not know.' There is some discrepancy in the affidavits of the eleven affiants as to the amount of money represented by the second computation; some put it at $19,369; others, at $19,545. All of the affiants who make mention of the fact that there were two computations give about the same reason, in effect, for having the second, i. e., that, from the start, nine jurors favored a ver *150 diet for plaintiff and three favored a verdict for defendant; that in the first computation all voted and the three who favored defendant voted for little or nothing and the result, something over $14,000, was too small to suit the most of the jurors and that the most of them favored another computation (to which all assented), with the suggestion by some that those favoring plaintiff materially increase the amount voted for, so as to overcome the negative effect of the three who voted for little or nothing. Some of the affiants depose that at each computation twelve jurors voted (something or nothing) and that the aggregate was divided by twelve. Others depose that that was done upon the first computation but that, upon the second, only the nine favoring plaintiff voted and that the divisor, the second time, was nine. All of the affiants agree that, from the start, the jury stood nine to three for plaintiff and that several ballots were taken on the amount to be awarded and on each there was wide divergence in amounts voted and no agreement. The affidavits differ as to why mathematical computations were then resorted to, the reasons given and. the understanding in regard thereto.

All of the affidavits filed by defendant declare that it was done “in order to determine the amount of damages to be awarded the plaintiff”; that it was understood that “the quotient so obtained should be the amount of damages and written into the form of verdict”; that “said suggestion was accepted and agreed to.”

All of the affidavits filed by plaintiff declare it was done “in order to determine the views of the jurors as to the average amount of damages”; that it “was purely informal, for the purpose of ascertaining the average sense of the jury as to the amount of damages to be awarded and as a basis for further consideration; that there was no agreement in advance of the making of said calculations, or either of them, to return a verdict for the quotient so found or to abide by the result of said calculations.” They further declare that, *151 after the second calculation, “further discussion of the merits of the case was had by the jury.” None of the affidavits filed by defendant say anything about “further discussion” but recite that the verdict followed immediately upon the result of the second calculation.

Whether the result of the last calculation was $19,369, as claimed by some, or $19,545, as claimed by others, different accounts are given as to how and -why the verdict was returned for $20,000. Defendant’s affidavits, after narrating the second calculation and its result, recite “that it ivas then suggested by one or more of the jurors that the amount be made an even $20,000, to which the nine jurors in favor of awarding plaintiff damages consented.” Four of plaintiff’s affidavits recite that, after the second calculation and further discussion, “it was unanimously agreed by the nine jurors who favored plaintiff that a verdict of $20,000 should be awarded the plaintiff.” The other two of plaintiff’s affidavits declare that, after the second calculation and further discussion, a ballot was taken, which was unanimous, as to the nine favoring plaintiff, for $20,000. There are some other details of the eleven affidavits, under discussion, not necessary to review.

Defendant first filed its five affidavits, above discussed; plaintiff then filed her six. Thereafter, defendant filed two supplemental affidavits, in reply. Those tivo we have not heretofore mentioned. They were made by Jurors Goodhue and Bickert, two of the five -whose affidavits were first filed by defendant and two of the nine jurors favoring plaintiff and joining in the verdict. Each of those two affidavits declares “that affiant consented to said verdict and the amount stated therein, because of the previous agreement between the said nine jurors, of which affiant was one, to fix and determine the amount of plaintiff’s damages in the manner stated and set forth in affiant’s former affidavit herein.” The first affidavit of “each recited that there was, in advance, a binding agreement to obtain a quotient verdict and abide by it.

*152 Thus, we have five jurors swearing that the mathematical calculations were made upon an express agreement and understanding that a verdict be arrived at by that process and that they were made for that purpose: six jurors swearing that the calculations were made to determine the views of the jurors as to the average amount of damages; that it was informal and was only a basis for further consideration; that there was no agreement in advance to abide by the result; that, after the last calculation, there was further discussion: and two of the five first referred to swearing further that they consented to the verdict and the amount thereof because of the preceding agreement (which they assert existed) of the nine jurors favorable to plaintiff, of whom they were two, to arrive at a verdict by mathematical process and to abide by it.

We do not question the motives of any of the deposing jurors. We concede all are men of honesty and veracity. That is a proper presumption and it should prevail until and unless there is something so strong and persuasive as to overthrow the presumption. We discern nothing of that character. We concede that, in all probability, each of the deposing jurors narrates truthfully, to the best of his ability, his recollection of events and understandings. That being the case, it is evident different jurors had different understandings of matters — doubtless honest differences of understanding. Evidently, five jurors understood there was an agreement to arrive at a quotient verdict and abide by it; evidently, six other jurors understood no such agreement but thought they were merely experimenting, as a basis for further consideration. It is evident that two of the five jurors first mentioned not only understood they were arriving at a quotient verdict but they swear their reason for assenting to it and the amount of it was because they had pledged themselves in advance so to do and felt bound by the pledge. Nobody contradicts those two as to their understanding and motives. In such matters, each person can swear only for *153 himself, as to what is or was in his mind, what is or was his understanding; nobody else can tell what is or was in his mind. “He, of course, could not testify as to the cognovit of the other jurors.” (International Agr. Corp. v. Abercrombie, 184 Ala. 244, 49 L. R. A. (n. s.) 415, 63 South. 549.)

Now, we concede that, as to the six jurors who made affidavits for plaintiff, all understood no agreement effective and all voted their free sentiments. For the sake of liberality, we will concede that three of the five jurors who made affidavits for defendant understood there was some sort of agreement but did not feel bound by it and may have voted their free sentiments. We accept as true the sworn word of the other two, Goodhue and Bickert, of the five jurors who made affidavits for defendant, to the effect that they understood there was a binding agreement and that because thereof they assented to the verdict. We will concede that all others favoring plaintiff voted their free sentiments, either because they were not aware of any agreement or did not feel bound by any. Nine jurors, including Goodhue and Bickert, concurred in the verdict. Of the nine, two concurred under compulsion of a previous binding agreement, as they undei’stood. That reduces the number of jurors concurring in a lawful manner to seven, one less than the number required to return a lawful verdict. The law says eight jurors must concur, freely and without compulsion, as a matter of honest judgment, in order to return a verdict. Here, at most, there were seven so doing. Will anybody say that was a lawful verdict? We say not.

The case of Wright v. Union Pac. R. Co., supra, is in point. There the jury consisted of eight jurors; six were required to concur, in order to return a verdict. Six concurred in a verdict for plaintiff. Defendant moved for a new trial on the ground that the verdict was obtained by resort to chance. It was established that mathematical calculation was made as in the case at bar. The quotient result was made the amount of the verdict. Five of the concurring jurors made *154 affidavits that there was no agreement that the quotient should be the amount of the verdict; that it was understood that no juror should feel bound by the result but that all jurors should feel free to advocate any amount. The other three jurors deposed that it was agreed and understood that the amount so arrived at should be the amount of the verdict. There was there, as in the case at bar, a sharp division; only there it was five to three that there was no binding agreement; here, six against five. In that case, one of the six concurring in the verdict was one of the three jurors deposing there was a binding agreement. That reduced the number of jurors freely concurring in the verdict to five, one less than required by law. The supreme court held it not a legal verdict and ordered a new trial. It would be hard, we think, to find two cases more analogous on fact and principle, involving any question, than that ease and the one at bar.

It may be said that this reasoning puts it in the power of two jurors, even one juror, by making false affidavits or affidavit, to invalidate any verdict rendered by a close vote, exactly the required number or one more. Yet, if two jurors should depose that there was an agreement for a quotient verdict and that it was carried out and ten other jurors should swear that such a thing was not mentioned; that there was no mathematical calculation — no such thing suggested or done; we dare say no trial or appellate court would hold there was a quotient verdict. It would be regarded, doubtless, as a case of palpable perjury of the two jurors. Here it is different. By the affidavits of eleven jurors (all those heard from) it is established that there were quotient calculations; that a quotient result was 'obtained; that, approximately, it was near the amount of the verdict. There are differences of understanding, however, about motives, objects, agreements; quite natural. Everything indicates that the two jurors who swear they assented because of being bound by a previous pledge so understood the situation.

*155 Counsel for plaintiff urge several objections to the contention that the verdict is the result of resorting to quotient process. One is that two mathematical calculations were made, the second resulting in an amount something near the amount of the verdict and the first, in a sum much less. They contend that whatever the agreement was it was with relation to the first calculation and was then abandoned. The affidavits do not so show. They show the two calculations were intimately related and were based on the same understanding, whatever it was, even though comprehended differently by different jurors. The two calculations were one connected transaction. The second was merely a continuation of what the jurors started out to do. Moreover, Jurors Goodhue and Rickert base their understanding and conduct wholly on the second calculation.

Another objection urged is that, while the quotient of the second calculation was $19,369, as deposed by some jurors, or $19,545, as deposed by others, the verdict was for $20,000, not, in any event, the quotient sum. It is generally held that, if the amount of the verdict is not exactly the quotient sum but is adopted as a result of the quotient process and grows out of the latter, the verdict is invalid. Authorities hold that a slight discrepancy is of no consequence, as affecting the invalidity of the verdict. (20 R. C. L. 244.) While authorities so hold about a slight discrepancy, there is no well-defined rule about what discrepancy is slight nor how much the- discrepancy may be and the verdict still be invalid. Much seems to rest on the basic facts that mathematical calculation was resorted to and led up to the fixing of the amount of the verdict, even though not the same as the quotient. Each ease seems to depend on its own facts and their relation to each other. Motives of jurors, as disclosed, are given much weight. Frequently the motive in agreeing upon a sum different from the quotient sum is to make it “even money” but that is held not a good reason.

*156 We have examined a number of eases in which there were variations, to make the amount “even money” or “round numbers,” and in which the verdicts, nevertheless, were held invalid; no variation was so large as in the case at bar but no rule is declared as to the extent to which a variation may go without validating the verdict. In McDonnell v. Pescadero & S. M. Stage Co., 120 Cal. 476, 52 Pac. 725, the verdict for plaintiff was for $2,000. An affidavit of defendant’s counsel, made on information and belief, asserted that the quotient process was used, upon an agreement to abide by it, and resulted in a quotient of $1,891.66 and that the jurors, in order to' make it an “ even sum, ’ ’ made the verdict $2,000. This was positively denied by the affidavits of two jurors, in which they declared there was no previous agreement and that the sum of $2,000 was arrived at as the result of discussion and deliberation. The court said: “If we were at liberty to accept the affidavit of defendant’s counsel and reject the affidavits of the two jurymen in support of the verdict, we would unhesitatingly decide that the verdict was reached by such a ‘resort to the determination of chance’ as would be good ground for a new trial.” Thus, no objection was made by the court to the discrepancy of figures, far greater relatively than in the case at bar, but the court held, the assertion of defendant’s counsel not sustained. Had it been sustained, the discrepancy would have been of no consequence. In Galveston, H. & S. A. Ry. Co. v. Brassell, supra, the discrepancy was not large but the court, in granting a new trial, condemned the actions of some of the jurors, who, as disclosed, agreed to return a verdict for an amount different from the amounts they thought right and just.

There is evidence of the same thing in the ease at bar and that is one of the worst features of it. Nine jurors declare in their affidavits there were two quotient calculations. They assert that upon the first each juror voted for what he thought just. It was a composite expression of the judgment of the jury and resulted in a quotient of about $14,000. The ma *157 jority thought that not enough. Another calculation was decided upon, to make it more. Three of the jurors depose that it was suggested then and agreed to that, in order to overcome the negative effect of three other jurors (opposed to plaintiff) who had voted for little or nothing, those favoring plaintiff “raise the ante — that is to say, increase the amount of the sum which each of the jurors would write down and vote for.” Thereupon, a second calculation resulted in a quotient of over $19,000. There is no reasonable doubt that those three jurors and little doubt that others, as well, on the second calculation voted for much more than they .thought just. It is inevitable, too, that some, in assenting to $20,000, must have assented to a much greater sum than they thought just. The law does not approve of such methods. Five of the eight jurors necessary to a lawful verdict and of the nine who concurred, depose they agreed to a verdict for $20,-000 to make it “even.” The two who depose that they concurred in the verdict because of a previous binding agreement to abide the result of the quotient process are among those who swear they consented to put the amount at $20,000, in order to make it “even.” There is in the record much evidence of misconduct and of methods which the law does not tolerate. We hold the verdict was not a lawful verdict and that the trial court rightly exercised its discretion in granting a new trial. Having come to that conclusion, it is not necessary to consider the other ground presented, the claim of excessive damages.

It is deplorable that juries, upon occasion, will use unlawful methods of arriving at verdicts, when it could be avoided; for the law requires fair verdicts, the result of honest judgment, not of devious schemes or chance methods. Juries should keep away from forbidden ground. We commend the practice of some district judges in habitually warning juries, by appropriate instructions, against quotient verdicts. *158 It would be wise if it were universally done. For the reasons herein given, the order appealed from is affirmed.

Affirmed.

Mr. Chief Justice Callaway and Associate Justices Stark, Matthews and Galen concur.