On remand, under the prior decision of this court, Thoms v. Andersen, 1975, 89 S.D. 567, 235 N.W.2d 898, the trial court, considering secondary evidence as mandated by this court, entered a decision declaring respondent, Donna Thoms, the winner of the election. Appellant has challenged the admissibility of the official *624canvass as evidence upon which the trial court’s decision was, at least in part, predicated. We affirm.
This court in Thoms v. Andersen, supra, affirmed the holding of the trial court that the ballots in precincts 3-4 and 10-2 had been improperly kept by the election officials and that the recount board properly did not count the absentee ballots from those two precincts. This court relied, in part, upon an early decision of the North Dakota Court, i.e., Howser v. Pepper, 1899, 8 N.D. 484, 79 N.W. 1018, see Thoms v. Andersen, supra, but did not delve into the reasons for the substitution of the official canvass results in those two precincts for a count of the ballots which were held to be tainted as evidence. The North Dakota Court in Howser v. Pepper, supra, set forth the best evidence rule as it relates to ballots in the following language:
“It may be stated, as a general principle, that the original ballots cast are the best evidence from which to determine the result of an election. The difficulty does not consist in the rule, but is found in the preliminary question as to whether the ballots offered to impeach the returns are the original ballots, and in the same condition as when cast by the electors. The effect of the original ballots as evidence, and the degree of care required to continue them as controlling evidence, is stated by Judge Brewer in Hudson v. Solomon, 19 Kan. 177, in the following language: ‘First. As between the ballots cast at an election and a canvass of these ballots by the election officers, the former are primary, — the controlling evidence. Second. In order to continue the ballots controlling as evidence, it must appear that they have been preserved in the manner and by the officers prescribed by the statute, and that, while in such custody, they have not been so exposed to the reach of unauthorized persons as to afford a reasonable probability of their having been changed or tampered with.’ ” 79 N.W. at 1019.
In keeping with the prior opinion of this court, respondent offered as evidence the official canvass which was marked as Ex*625hibit #7. This canvass was duly filed and was complete in all respects save that the signatures of the board were not certified to and the seal of the county auditor was not affixed as required by SDCL 12-20-38.
Because of this defect, appellant claims the exhibit was improperly received in evidence and, further, even if properly admitted, the canvass does not prove the results of the vote in the two precincts in question since the canvass shows the total vote for each candidate in each precinct without a breakdown of the machine vote and absentee vote. The latter contention, as well as appellant’s further contention that the trial court improperly used a stipulation of counsel relative to the number of absentee votes cast for each candidate in precinct 3-4 in arriving at its decision, we find unnecessary to discuss by reason of our prior decision and our holding herein.
In an early discussion of the best evidence rule, the territorial court of Dakota Territory acknowledged that
“The principle of the rule requiring the best or highest evidence, is founded on the presumption that there is something in the better evidence which is withheld, adverse to the party resorting to inferior or secondary evidence. The general effect of the rule is to prevent fraud, and to induce parties to bring before juries the kind of evidence least calculated to mislead or perplex them. And the reason of the rule limits the extent of its application. Consequently, it does not operate where the law itself obviates the presumption of fraud, which would otherwise arise. * * * ” McCall v. United States, 1876, 1 Dak. 307, 46 N.W. 608.
Considering that the county auditor, who was responsible for the care and safekeeping of the ballots returned to him, had taken such little care in the performance of that duty as to render the ballots inadmissible as evidence, it would appear, in retrospect, that no confidence should have been placed by this court in that officer’s fulfilling the duties imposed by law for the preparation and preservation of the official canvass.
*626Whether the auditor’s failure to certify the signatures and affix his seal to the canvass as required by law is purely a ministerial act or is mandatory, see Smith v. Lawrence, 1891, 2 S.D. 185, 49 N.W. 7, so as to affect the admissibility of the exhibit under either SDCL 19-7-9 or 19-7-11, we need not determine.
The best evidence rule requires that secondary evidence must be looked to where the primary evidence has lost credence. However, a fraud may not be perpetrated in the name of the rule and the official canvass is not the only secondary evidence that may be considered.
As stated in Howser v. Pepper, supra, and quoted with approval by this court in Thoms v. Andersen, supra, where ballot boxes have been so improperly supervised that the ballots themselves have lost “their probative weight as the best evidence * * * they cannot, therefore, be allowed to impeach the canvass of the votes by the precinct officers * * 235 N.W.2d at 900. Tally sheets and the official return sheets from precincts 3-4 and 10-2, which constitute official returns for each precinct, SDCL 12-20-16, are the “canvass of the votes by the precinct officers” within the meaning of the quoted language from Howser v. Pepper, supra.
The precinct canvasses for both precincts were, without objection, before the trial court as official records. RCP Rule 44. These exhibits are prima facie evidence of the facts stated therein, SDCL 19-7-9, and they clearly establish that in precinct 3-4 the absentee votes cast were 42 for appellant, Andersen, and 98 for respondent, Thoms. They also establish that in precinct 10-2, 33 were absentee votes cast for appellant and 43 for respondent. Those votes, when added together with the unchallenged results of the recount board declaring the vote of 18,705 for appellant and 18,688 for respondent, show respondent to be the winner by a vote of 18,829 to appellant’s 18,780.
Affirmed.
WINANS and WOLLMAN, JJ„ BIEGELMEIER, Retired J„ and FOSHEIM, Circuit Judge, concur. *627BIEGELMEIER, Retired J„ sitting for DUNN, C. J., disqualified. FOSHEIM, Circuit Judge, sitting by assignment.