Methods of the Second Class.

The simplest method of the second class is the French method of double elections. In this method each elector has one vote, as in the single vote method, already described. If, however, no candidate obtain an absolute majority of the votes polled, a second election is held. For this second election only the two candidates who obtained the largest number of votes at the first election can be candidates. The result is that the successful candidate is returned by an absolute majority of those who vote at the second election, so that it would appear, at first sight, that the successful candidate represents the views of a majority of the electors. We must not lose sight, however, of two facts, first, that all the electors who vote at the first election may not vote at the second election; second, that those who do so vote merely have to choose between the two remaining candidates, and that, consequently, they may not be repre- page 6 seated in any sense by the candidate they vote for; they may merely be in the position of having a choice of evils.

This plan has frequently been proposed for adoption in England, and quite recently it has been proposed by more than one speaker in the Legislative Assembly of Victoria. The method is indeed a great improvement on the present system of single voting, and if the election be merely a party contest, and neither side runs more than two candidates, the result cannot be wrong. But if these conditions be not satisfied, the method may easily lead to an erroneous result. The method may be used whatever be the number of candidates; but it is sufficient to show that it is erroneous in the case of three candidates only. This is at once done by a further consideration of the example already given in discussing the single vote method. For in that example C is at the bottom of the poll, and, according to the present system, he is rejected, and a second election is held to decide between A and B, because no one has an absolute majority at the first election. The result of the second election is, for A, five votes; for B, seven votes; so that B wins. In order to show that this result may be erroneous it is only necessary to suppose that the live electors who voted for A prefer C to B. For then, if the question

That C is to be preferred to B

was put to the whole body of electors, it would be carried by a majority of four. Now we have already seen that the question

That C is to be preferred to A

would be carried by a majority of two. Hence, then, this method leads to the rejection of a candidate who is declared by a majority of the electors to be superior to each of the other candidates. This method, then, clearly violates the condition that the result must not be contrary to the wishes of the majority.

We may consider this example from a slightly different point of view. In discussing it under the single vote method, the important result arrived at was that A was inferior to each of the other candidates, and, therefore, ought to be at the bottom of the poll, instead of being at the top, as he was, in consequence of his being opposed by two good men, B and G. Thus, instead of excluding C, as in the French method, A is the one who ought to be excluded. Having arrived at the result that A is to be excluded, the whole of the electors have now a right to decide between B page 7 and C. On putting this question to the issue, we find that C is preferred by the electors.

We see, then, that the French method may lead to error through throwing out the best man at the first election. And this is the only way in which it can err; for if there be a best man, and he survive the ordeal of the first election, he must win at the second, seeing that he is, in the opinion of the electors, better than each of his competitors.

Comparing the French method with the single vote method, we see that in the case of three candidates the worst candidate may be returned by the single vote method, but that it would be impossible for such a result to be brought about by the French method. By that method we are at least sure of getting the second best man, if we fail to get the best.

There is, however, a grave practical objection to this method. It is that a second polling may be necessary. This is of great importance; for in the case where the number of electors is large, as in a political election, great expense has to be incurred, not only by the authorities in providing the necessary machinery, but also by the electors themselves in coming to the poll again. Besides this, the excitement of the election is kept up much longer than it would be if the whole matter could be settled by a single polling. There can, I think, be little doubt that this objection has been one of the chief obstacles with which the advocates of this method have had to contend. Accordingly, we find that the single vote method is employed, as a rule, in those cases in which there are some hundreds of electors, and it would be inconvenient to hold a second election. On the other hand, when the number of electors is small, so that they can all meet together, and remain till a second or third election has been held, the number of candidates is generally reduced to two by means of a preliminary ballot or ballots. This very fact shows that the defects of the single vote method are recognised, because in those cases in which it is considered to be practicable to do so a preliminary election is held, so as to try to avoid the glaring defect of the single vote method—that is, to avoid returning a candidate who is acceptable to a small section only of the electors. It is a mistake, however, to suppose that it if not practicable to hold one or more preliminary elector when the number of electors is large. It is general thought that in order to do so a fresh set of voting [unclear: pap] page 8 must be used for the second election, and that this second election cannot be held till the result of the first is known, so that the electors have the expense and trouble of going to the poll a second time. This, at all events, appears to be the practice in France, Germany, and Italy. This, however, is not necessary; for, by a very simple expedient, any number of preliminary elections, on any plan whatever, may be held by means of a single set of voting papers, and without troubling the electors to vote more than once. The expedient is to require each elector to indicate his order of preference amongst all the candidates. Once get this information from the electors, and we can tell how any elector will vote on any question that may be put as to the merits of the candidates. It is here assumed that an elector will not change his opinion during the course of the election. This expedient of making each elector indicate his order of preference amongst all the candidates is necessary in order to carry out Borda's method, which has been described above; indeed, it was suggested by Borda himself. But Borda does not appear to have noticed that it might be made use of for a series of elections without requiring the electors to vote again; this appears to have been first pointed out by Condorcet. The idea of a preferential or comparative voting paper is one of the fundamental ones in Hare's system of proportional representation. We are not concerned with this subject here, as the only question under consideration is that of filling a single vacancy. It is, however, worthy of notice that the preferential voting paper which is such an important feature in Hare's system, is of such old origin, and that it was suggested by Condorcet as a means of filling several vacancies, which is the very question considered by Hare. The method of Condorcet, however, is quite different to that of Hare.

If the expedient here described were adopted, the French system would be free from the practical objection which has been indicated. It would still, however, be open to the objection that the result of the election might be contrary to the views of the electors. Notwithstanding this, the method would be a good practical one for elections on a large scale; it would be very suitable for party contests, and if neither side ran too many candidates, the result could not be wrong. The method, however, would be altogether unsuitable if there were three distinct parties to the contest. Under any circumstances, however, the method would be page 9 very little more complicated than the present system of single voting, and it would give much better results. If, however, it be considered desirable to reform the present electoral system so far as to introduce this French system of double elections, it would be as well to at once adopt the method of Ware, described below. This is the same, in the case of three candidates, as the French method, but in other cases it is a trifle longer. No difference whatever would be required in the method of voting, but only a little more labour on the part of the returning officer. The results of this method would be much more trustworthy than those of the French method.