In the 1963 Democratic primary election for male leader of the First Assembly District, Part A, New York County, the rival candidates were appellant DeSapio and respondent Koch. The official canvass recorded the election of Koch by 41 votes out of a total of more than 9,000 cast. DeSapio has brought this proceeding under subdivision 2 of section 330 of the Election Law, seeking an order directing the holding of a new primary election for the party position involved. After extended and comprehensive hearings, Special Term dismissed DeSapio’s petition.
A new type of voting machine was used for the first time in this primary election. It was equipped with a single public counter, which reflected the aggregate of the votes cast in both the Democratic and Republican primaries. Before admitting a voter inside the curtain shielding the machine from view, an inspector of election regulated a device known as a party selector. This engaged the machine so that a voter could vote only in the primary of his party enrollment.
Two identifiable groups of votes are contested specifically by DeSapio. It is undisputed that 45 persons voted in the Democratic primary who did not sign the registration poll records, as provided by the Election Law (§§ 201, 412, 413) and the State Constitution (art. II, § 7). He also contends that 24 additional persons voted illegally because they were not enrolled Democrats or did not reside within the district. The Justice at Special Term sustained objections to 15 votes in *22the first category, and 20 in the second — 35 in all. We are in substantial agreement with his painstaking and well-reasoned holdings on objections to specified votes; at the most three more objections might have been sustained.
This was a close election. To repeat, the official canvass reported Koch the winner by 41 votes. Sixty-five specific votes were contested by DeSapio. Special Term sustained 35 of his objections. This does not mean, of course, that Koch’s margin of victory was thereby cut to 6 votes, since no one except the voters themselves knows for whom the 35 invalid votes were cast (Matter of Badillo v. Santangelo, 15 A D 2d 341).
Close elections usually leave in their wake nagging suspicions that perhaps the true choice of the electorate was not declared the winner. But all elections do not result in thumping pluralities that give reassuring evidence of the clear-cut mandate of the People; and there is no law in this State providing that elections of a specified closeness must be rerun. The margin of victory, no matter how narrow, in and of itself cannot justify upsetting an election (Matter of McGuinness v. DeSapio, 9 A D 2d 65). There must at least be a showing that would justify a reasonable belief that the challenged irregularity may have accounted for the victor’s plurality. Absent such a showing there must be a finality to closely fought elections just as in closely fought lawsuits.
When an election is as close as this one was, the first impulse of the defeated candidate is to try to bowl over the slight difference in recorded votes through an exercise in mathematics. If the number of invalid votes were three times the 35 found by Special Term, that could warrant setting aside this election; but on simple arithmetic alone, 35 invalid votes cannot erase a plurality of 41 votes. On the other hand, it cannot be gainsaid that the narrow margin of victory is a most influential element in the complex of factors that must be considered in determining whether the recorded result truly reflected who was rightfully elected.
There will be elections in which few if any votes can be identified and invalidated as conclusively as those challenged in this proceeding. Such elections may be conducted so badly that even though illegality of specific votes cannot be attributed to the misconduct, still it must be found that the resultant mischief held such potential for changing the result that every dictate of fairness and protection of the voters’ franchise demands a new election. In such a situation one must look to the conduct and climate of the election as a whole.
*23For example, it seems to me that where widespread fraud is proven, the specific disclosure of a clutch of illegal votes aggregating a substantial proportion of the winning plurality might support a conclusion that undiscovered fraud accounted for the balance of that plurality (cf. Matter of Bloom v. Power, 21 Misc 2d 885, 891, 892, affd. 9 A D 2d 626, affd. 6 N Y 2d 1001; Matter of Burns [Sullivan], 303 N. Y. 601; Matter of Haas v. Costigan, 14 A D 2d 809, affd. 10 N Y 2d 889; Matter of Hooper v. Power, 17 A D 2d 816, affd. 12 N Y 2d 764; Matter of Friedman, 238 App. Div. 341, 344; Matter of Weisberger v. Cohen, 260 App. Div. 392). In this proceeding, however, no claim of fraud has ever been asserted.
Or, in a close contest, the inefficiency or carelessness of the persons conducting the election may cast sufficient doubt on the result to warrant a new election. In this respect an election contest again is not unlike a contest in a court of law. Either may be set aside for prejudicial errors committed by the officials who conduct them; but their responsibility should be measured by the fact that the interested parties are represented by lawyers or election district watchers. Of necessity, lawsuits and elections are to some extent regarded as adversary actions, to be fought hard and cleanly, it is true, but within a dominant self-help philosophy (Matter of McGuinness v. DeSapio, supra, p. 73).
As would be expected in a bitter primary fight embracing 43 election districts, with over 11,000 persons voting in the two-party contests, and more than 9,000 in the Democratic primary, the election officials made mistakes. The high ratio of identified illegal votes to Koch’s plurality reduces, of course, his margin for additional error of a more problematical nature. We address ourselves, therefore, to the question as to whether the entire pattern of the election would reasonably be found to have so affected the result as to require a new election, according to the foregoing guidelines.
As previously stated, 45 persons who did not sign the registration cards were permitted to vote. This indicates neglect of duty by the election officials; but a contained kind of neglect that could not affect more than the 45 votes specifically identified as cast by nonsigners. Had other nonsigners been permitted to vote, those irregularities would inexorably have been revealed by the comparisons made between the signed registration cards and the public counter tally. We may not speculate that because election officials were revealed to be negligent in one circumscribed area they must have been negligent to *24the detriment of DeSapio in other areas unknown and unmentioned.
The same must be said of a similarly limited group of irregularities involving the 24 persons who voted illegally in the Democratic primary because they were not enrolled in the Democratic party, or because they did not reside within the district on primary day and were therefore ineligible to vote.
There was only one contest in the Republican primary — for the county-wide office of Councilman-at-Large. This engendered comparatively little of the local fierceness and bitterness of the Democratic primary. Contrary to the experience in the Democratic primary, in which it will be recalled 45 more persons voted than signed the registration records, 23 less persons voted in the Republican primary than signed the records. DeSapio claims it must be inferred that these 23 Republicans voted illegally in the Democratic primary. In such event, however, these votes would be merged into the previously mentioned 45 votes in the Democratic primary that were in excess of the signatures in the registration records. Special Term properly rejected this contention as too speculative, noting that some of the discrepancy might have resulted from voters leaving the polling place without voting.
We now consider another group of irregularities. They involve a large block of votes, but there is just no basis for estimating, with any degree of reliability, that invalid votes resulted from these omissions of the election officials.
The Board of Elections, in accordance with subdivision 7 of section 242-a of the Election Law, furnished the local Boards of Inspectors with public counter cards. These cards are designed to provide a link of communication between the inspectors who procure the signature of the voter to the registration record and the inspectors in charge of the voting booth, so that a person may vote only in the primary of the party in which he is enrolled. The registration book inspectors were required to insert the name and public counter number of each voter on these cards, after he had signed the registration record. Cards colored green were given voters eligible to vote in the Democratic primary and cherry-colored ones for the Republican primary. The voting booth inspectors collected the public counter cards from the voters and adjusted the party selector device so that they could vote only in the appropriate primary. The Boards of Inspectors were instructed to return the public counter cards to the Board of Elections at the close of primary day. In four election districts the Boards of Inspectors failed to return the public counter cards and they *25were evidently destroyed or thrown away. There were abont 1,100 such cards — representing abont 10% of the total primary vote.
DeSapio argues that a number of ineligible persons may have voted in the Democratic primary in the four election districts that did not return the public counter cards. The returns for those districts refute this contention. Nine hundred seventy-seven persons signed the registration records who were eligible to vote in the Democratic primary, and 941 actually voted in that primary. One hundred forty-one entitled to vote in the Republican primary signed the registration records and 139 did in fact vote in that primary. These figures give scant comfort to the appellant in this area. The differential of 36 eligible Democrats who refrained from voting in these four districts is unexplained; and in any event, it is difficult to see how the availability of the green cards would have shed any light on the discrepancy.
Finally, Special Term sustained the validity of 27 votes cast in the Democratic primary by voters who did not sign the registration records. These 27 votes were of course included in the afore-mentioned excess of 45 votes in the Democratic primary over and above the number of eligible voters who signed the registration records. Special Term overruled DeSapio’s objections to the 27 disputed votes, essentially upon impressive testimony of persons that they had voted without signing the registration records and upon production of green public counter cards bearing their names. Had these 27 votes been held illegal, DeSapio would have established the invalidity of at least 62, possibly 65 specific votes in the Democratic primary election.
But even if all these votes had been held illegal, 53 of them must have been cast for Koch to have changed the result of the election. In an election so close and concededly free from fraud, it is highly unlikely that Koch received 53 out of those 65 votes (Matter of Badillo v. Santangelo, 15 AD 2d 341, supra).
Matter of Branagan v. Todd (19 A D 2d 337, affd. 13 N Y 2d 888), as we construe it, did not turn starkly on the illegality of the 47 votes cast on one machine in a Democratic primary in excess of the number of Democratic voters who voted in that election district. Rather, the irregularities were found to have resulted from “ the improper functioning of the voting machine ” (p. 339), and by implication related to all the votes registered by that machine. Since the petitioner in that case enjoyed a clear lead on the computation of votes cast in the remainder of the city which was offset by the tabulation of *26votes on the defective machine, no mathematical exercise was required to hold that if the machine had functioned properly there was a reasonable likelihood the petitioner’s lead might not have been destroyed.
DeSapio’s major thrust upon this appeal is that the State Constitution and implementing legislation mandate as a condition precedent to participation in an election that the person presenting himself to vote sign his name on the registration poll record; and that the court cannot relieve a voter of so unequivocal a requirement to sign the registration record. In rejecting this contention Special Term relied upon the 1938 holding of the Appellate Division, Second Department, in Matter of Moritt v. Cohen (255 App. Div. 804, 805), which held among other things that 1 ‘ the oversight of the inspectors of election in not requiring them to sign should not be visited upon voters otherwise qualified.” All of us except Mr. Justice Rabin are disposed to follow this holding, although we are mindful of the fact that in affirming the Court of Appeals evidently found it unnecessary to reach this aspect of the Appellate Division’s decision.
In summary, then, the specifically voided votes are not in themselves sufficient in number to warrant setting aside the election, and the evidence of other irregularities does not in our opinion offer the potential for invalidating additional votes. Special Term’s disposition should therefore be affirmed, with costs.