CHAPTER VII
THE SINGLE TRANSFERABLE VOTE
"The law regulating the form of voting may be thus expressed. Every vote shall be given on a document setting forth the name of the candidate for whom it is given; and if the vote be intended, in the events provided for by this Act, to be transferred to any other candidate, or candidates, then the names of such other candidate, or candidates, must be added in numerical order."—Thomas Hare, The Election of Representatives (Fourth edition, 1873)
The single transferable vote was the distinguishing characteristic of the scheme of electoral reform proposed by Hare in 1857, but it was associated with the proposal to treat the whole kingdom as a single constituency. The later advocates of this new method of voting have recommended its application to constituencies of more moderate size, such as counties and large towns, and in this form the system has found a more ready acceptance and has been used with success in parliamentary elections.
Its present application.]
The first application of the single transferable vote took place in Denmark[1] in 1855, and it is still being used under the Constitution of 1867 in the election of members of the Danish Upper House. It is also used, as provided by the South Africa Act of 1909, in the elections of the Senate of the United Parliament and in the election of the Executive Committees of the Provincial Councils. In each of these cases the electorates are small, and the electors possess special qualifications. The Danish Upper House is elected in two stages, the transferable vote being used only in the final stage in which electors of the second degree alone take part. In South Africa the members of the first Senate were elected by members of the local parliaments of the several Colonies,[2] and the Executive Committees of the Provincial Councils by members of the Councils. The system has, however, been subjected to the test of popular parliamentary elections in Tasmania and of municipal elections in Pretoria and Johannesburg.
Ever since the publication of Hare's scheme, proposals for proportional representation have been associated in English-speaking countries with the idea of a transferable vote. Hare's proposals were warmly endorsed by John Stuart Mill first in Representative Government, and again in a memorable speech delivered in the House of Commons on 30 May 1867, when he moved an amendment to the Electoral Reform Bill.[3] Mill's amendment was defeated, but he retained to the full his faith in the great value and need of the improved method of voting, as the following passage from his Autobiography shows: "This great discovery," said he, "for it is no less, in the political art, inspired me, as I believe it has inspired all thoughtful persons who have adopted it, with new and more sanguine hopes respecting the prospects of human Society, by freeing the form of political institutions towards which the whole civilized world is manifestly and irresistibly tending from the chief part of what seemed to qualify and render doubtful its ultimate benefits. … I can understand that persons, otherwise intelligent, should, for want of sufficient examination, be repelled from Mr. Hare's plan by what they think the complex nature of its machinery. But any one who does not feel the want which the scheme is intended to supply; any one who throws it over as a mere theoretical subtlety or crochet, tending to no valuable purpose and unworthy of the attention of practical men, may be pronounced an incompetent statesman, unequal to the politics of the future."[4]
An English movement.]
The English advocates of proportional representation who have succeeded Mill have equally favoured the single transferable vote. This system was embodied in the Bill introduced into the House of Commons in 1872 by Mr. Walter Morrison, Mr. Auberon Herbert, Mr. Henry Fawcett, and Mr. Thomas Hughes; it was advocated in the important debates which took place in the House of Commons in 1878 and 1879; and the Proportional Representation Society, founded in 1884 in view of the Electoral Reform Bill of that year, created, under the leadership of Sir John Lubbock and Mr. Leonard Courtney, a strong movement in its favour. Owing to the agreement between the leaders of the Liberal and Conservative parties in favour of single-member constituencies this movement had no immediate result. Since its revival in 1905 the Proportional Representation Society has continued to press the claims of the single transferable vote, and with some success. The practicability of the system was admitted by the Select Committee of the House of Lords appointed to examine the Municipal Representation Bill introduced into that House by Lord Courtney in 1907; the model elections organized by the Society in 1906, 1908, and 1910,[5] have to some extent familiarized the British public with its details; it found, as already mentioned, a place in the South African Constitution of 1909, whilst the Royal Commission on Electoral Systems reported in 1910 that "of schemes for producing proportional representation we think that the transferable vote would have the best chance of ultimate acceptance."
The system in brief.]
What then is the single transferable vote, and how does it help to secure a true representation of the electors? Its mechanism and advantages will best be understood by a comparison with the existing system. The city of Birmingham is at present divided into seven single-member constituencies, with the result that the majority in each of these constituencies secures a representative, while the minority in each case is unrepresented. Suppose there were in Birmingham 40,000 Unionist, 20,000 Liberal, and 10,000 Labour voters: it might easily happen that the Unionists would be in a majority in each of the seven divisions and, if so, the 40,000 Unionist electors would obtain the seven seats and the remaining 30,000 voters none. The transferable vote, as will presently appear, would enable these 70,000 citizens to group themselves into seven sections of equal size, each returning one member, so that there would be four Unionist groups returning four members, two Liberal groups returning two members and one Labour group returning one member; and this is the ideal representation of such a community.
Large constituencies.]
In order to achieve this result several changes in electoral mechanism are required. In the first place, Birmingham, instead of being divided into seven constituencies, must be polled as one constituency, otherwise the necessary grouping could not take place. This change is not in itself sufficient, because if Birmingham were polled as one constituency electing seven members, and if each elector could give, as with the "block" vote, one vote apiece to seven candidates, then the seven nominees of the majority would all receive a higher number of votes than the seven nominees of the minority. In the numerical case cited above, each Unionist candidate would command 40,000 votes, each Liberal 20,000, and each Labour candidate 10,000, and the largest party would win all the seats.
The single vote.]
It is therefore necessary, however many may be the number of members to be elected, to limit the voting power of each elector to one vote—hence the name "the single vote." An obvious result of this limitation is that if a group numbering 10,000 electors concentrates its support upon one man, then the group is certain of returning that candidate, because not more than six equally large groups can be formed out of the remaining electors. With open voting the grouping of electors could be arranged with comparative ease, for if more electors than were sufficient to constitute his group desired to vote for a particular candidate, those who arrived late at the poll could be asked to give their votes to another candidate, and so help to build up another group of the requisite size. Or, if a candidate was receiving so little support that he had no chance of election, the small group that had gathered round him could be disbanded and these electors, instead of having their votes wasted, could make their selection from among the other candidates available. In this way seven groups could be formed, each of which would obtain a representative.[6]
The vote made transferable.]
As, however, the ballot is secret and the result of the voting is not known until the close of the poll, some provision must be made to facilitate the equal grouping of the electors upon which fair representation depends. This will be made clear by an example. Were Mr. Joseph Chamberlain one of the Unionist candidates for Birmingham, the group of voters who would record their votes for him would probably considerably exceed the number required for his election. His Unionist colleagues might, in consequence, find themselves left without adequate support, and the party might fail to secure its fair share of the representation. In order to prevent a mischance of this kind the very simple device has been adopted of making the vote transferable. By this means the necessary accuracy in grouping is secured automatically.
How votes are transferred.]
The transferable vote enables the elector to instruct the returning officer to whom his vote is to be transferred in the event of his first favourite either receiving more support than he requires or receiving so little as to have no chance of election. Continuing the example already given, an elector who desired to vote for Mr. Chamberlain would place on the ballot paper the figure 1 against his name. If, in addition, he placed the figures 2, 3, &c. against the names of other candidates in the order of his choice, these figures would instruct the returning officer, in the event of Mr. Chamberlain obtaining more votes than were necessary to secure his election, as to whom the vote was to be transferred. The votes given to Mr. Chamberlain in excess of the number required for his election would thus be rendered effective. They would be used and not wasted. If, on the other hand, an elector had recorded his vote for a candidate who, after all excess votes had been transferred, was found to be at the bottom of the poll, the returning officer would similarly give effect to the wishes of the elector as recorded on the ballot paper by transferring the vote to the elector's second choice. Again the vote would not be wasted, but would be used in building up a group sufficiently large to merit representation.
The ideas which have led up to the single transferable vote are, therefore, of a simple character. Constituencies returning several members are formed. A representative is given to every group of electors which attains to a definite proportion of the whole, the proportion depending upon the number of members to be returned. If a candidate receives more votes than are sufficient, i.e. if too large a group is formed, the surplus votes are transferred. If, after all surplus votes have been transferred, there still remain more candidates than there are vacancies, the lowest candidate on the poll is eliminated from the contest, i.e. the smallest group is disbanded. The transfer of surplus votes and of votes recorded for the candidates lowest on the poll are all carried out in accordance with the wishes of the electors as indicated by them on the ballot paper at the time of the poll. The proportionate representation of all the electors is secured; each party obtains the number of members to which it is entitled.
The Quota.
A few questions will at once occur to the reader as to the application of these simple rules. How is the number of votes required for success to be determined? In what way are the surplus votes to be distributed? What is the order in which the elimination of unsuccessful candidates shall proceed? The number of votes necessary to secure the election of a candidate is called the "quota." At first sight it would seem that this number should be ascertained, as suggested in the preceding paragraphs, by dividing the number of votes by the number of vacancies. But a smaller proportion is sufficient. Thus, in a single-member constituency a candidate has no need to poll all the votes; it is evident that if he polls more than a half he must be elected. No other candidate can equal him; the quota in this case is, therefore, one more than a half. So, in a two-member constituency the quota is one more than a third, for not more than two candidates can poll so much; in a three-member constituency, one more than a fourth, and so on. In a seven-member constituency, like that of Birmingham, the quota would be one more than an eighth. In general terms the quota is ascertained by dividing the votes polled by one more than the number of seats to be filled and adding one to the result.[7]
A simple case.
The processes involved in distributing the votes are described at some length in the account which appears further on in this chapter of the model election organized by the Proportional Representation Society in 1908, but the method of transferring votes and deciding the result of an election may be more easily understood from a simple case. Let us imagine there are six candidates for three seats, of whom A, B, C belong to one party and X, Y, Z to another. On the conclusion of the poll the ballot papers would be sorted into heaps, or files, corresponding to the names against which the figure 1 had been marked, and in this way the number of votes recorded for each candidate would be ascertained. Let us assume that the result of the sorting is as follows:—
A is marked 1 upon 1801 papers, and therefore has 1801 votes
B " 1 " 350 " " 350 "
C " 1 " 300 " " 300 "
X " 1 " 820 " " 820 "
Y " 1 " 500 " " 500 "
Z " 1 " 229 " " 229 "
—— ——
Total number of papers 4000 Total number of Votes 4000
As there are three seats the quota is one more than a fourth of the total of the votes polled. The total in this case is 4000, and the quota is therefore 1001.
A, having obtained more than the necessary quota of votes, is declared elected.
The transfer of surplus votes.
It will be seen that A has obtained nearly two quotas of votes, and his supporters, in the absence of any provision for the use of his surplus votes, would not obtain the full share of representation to which they are entitled. The next step is therefore to transfer A's surplus votes in accordance with the wishes of his supporters. These have indicated on the ballot papers to whom they desire their vote to be transferred. The different methods in which the transfer of votes can be carried out will be described, but for the present it may be assumed that the result of the operation was to transfer:
648 of the 800 surplus votes to B (a member of the same party as A) 132 " 800 " C (also a member of A's party) 20 " 800 " Z
The votes transferred to the several candidates are added to those already obtained by them as follows:—
Original Votes. Transferred Votes. Total.
B 350 + 648 = 998
C 300 + 132 = 432
X 820 nil = 820
Y 500 nil = 500
Z 229 + 20 = 249
The elimination of the lowest unelected candidate.]
Had any candidate, as a result of the transfer of A's surplus votes, been raised above the quota he would have been declared elected and his surplus distributed in the manner just described. In this case no candidate, as the result of the transfer, has obtained the quota, and there are, therefore, no further surplus votes to distribute. There are, however, two vacancies still remaining unfilled, and the next operation is to distribute the voting papers of Z, who, being the lowest on the poll, is clearly out of the running. Z's papers are sorted, as in the previous process, according to the candidates who are marked by the voters as their next preferences, and it may be supposed that the result is as follows:—
B is marked as next preference on 20 papers
X " " 200 "
Y " " 29 "
These papers are then added to the heaps of the respective candidates, B, X, and Y, and, with these additions, the votes credited to each candidate may be shown thus:—
Previous Transfer of
Total. Z's Votes. Total.
B 998 + 20 = 1018
C 432 + nil. = 432
X 820 + 200 = 1020
Y 500 + 29 = 529
Since B and X, as a result of the distribution, each obtain a quota of votes, they are declared elected, and all the vacant seats now being filled, the election is at an end.
The result.
The candidates elected, A, B, and X, each represent a "quota" of voters. Each considerable section of the constituency is thus able to choose a representative, whilst the party to whom both A and B belong return two members, these candidates taken together having secured the support of two quotas of voters. The voters who failed to secure a representative, namely the supporters of C and Y, number less than a quota.
Different methods of transferring surplus votes.—The Hare Method.]
There are several methods by which surplus votes may be transferred. In the case imagined the simplest way to distribute A's surplus votes is to take the 800 papers last filed and to sort these papers according to the second preferences indicated thereon. This method, which was recommended by the advocates of proportional representation in the movement of 1884-85, is based upon that contained in Mr. Hare's proposals. It has, however, been objected that if some other 800 voting papers are taken the result may be different, and that in this way an element of chance is introduced. This objection is considered in detail in Appendix VI., and it will be sufficient to state here that, when large numbers of votes are dealt with and the papers are well mixed, this element of chance is negligible. But small as it is it can be eliminated by adopting more accurate methods of transferring the votes.
The Hare-Clark method
One of these more accurate methods was embodied in the Tasmanian Act of 1896, and also in the Municipal Representation Bill approved by the Select Committee of the House of Lords in 1907. It is known as the Hare-Clark system, its inception being due to Mr. Justice Clark, of Tasmania. With this method the surplus votes of any successful candidate are transferred to the unelected candidates in such a way that each unelected candidate marked as the voter's next preference on the successful candidate's papers receives a proportionate share of the surplus. Continuing with the illustration already given, the returning officer, instead of taking from A's heap the 800 papers last filed, takes the whole of A's heap and sorts all these papers according to the next preferences. Assume that the result is as follows:—
B is marked 2 on….. ……………… ..1296 papers
C " 2 on……… ………….. .. 264 "
Z " 2 on…………. ………. .. 40 "
Total papers showing second preferences .. 1600
Papers on which no further preferences are shown …201
Total of A's papers……………….. …1801
In this case there are 800 surplus votes, whilst there are in all 1600 papers on which next preferences have been marked. It is therefore clear that each of the candidates B, C, Z is entitled to receive one-half the papers on which his name has been marked as the next preference. Each of the three bundles of papers showing next preferences for B, C, Z are divided into two portions. One portion is transferred to the next preference, the other is retained for the purpose of constituting A's quota, in which is included the papers on which A's name is alone marked.
The complete operation is shown below:—
Candidate indicated as Number Number of Number of
next Preference. of next Papers Transferred Papers
Preferences. to the next Retained for
Preference. A's Quota.
B 1290 648 648
C 264 132 132
Z 40 20 20
—— —- —-
Total of next preferences 1600 800 800
Papers showing no further preference 201 — 201 —— —- ——
Totals 1801 800 1001
In this way each of the candidates B, C, and Z obtains in strict proportion that share of A's surplus to which he is entitled, and, so far as this operation is concerned, the element of chance is wholly eliminated.[8]
The papers selected for transfer, however, are those last filed in the process of sorting, and should it become necessary to transfer these papers a second time there would enter in this further distribution an element of chance which, as explained in the Appendix already referred to, is so trifling as to have no practical effect upon the result unless the number of electors is small as compared with the number of members to be elected.
The Gregory Method.
A third method, in which the element of chance is eliminated from every transfer, has been embodied in the Tasmanian Act of 1907. Whenever it is necessary to transfer surplus votes, the whole of the successful candidate's papers on which preferences are marked are transferred, but at a reduced value. In the example given the whole of A's papers on which next preferences had been marked for B, C, and Z would be carried forward to those candidates, but each paper would be transferred at the value of one-half, the remaining portion of the value of each paper having been used for the purpose of electing A. This method is known as the fractional, or Gregory, method of transfer, having been first suggested by Mr. J. B. Gregory of Melbourne, in 1880. The regulations for the conduct of elections contained in the Tasmanian Act are given in Appendix VIII.
The committee which investigated the working of this system as applied to the Tasmanian General Election of 1909, made a very valuable comparison between the rules contained in the Municipal Representation Bill[9] and the more exact rules of the Tasmanian Act. A fresh scrutiny, based on the rules of the Municipal Representation Bill, was made of all the ballot papers used in that election. It was found that in each district the same candidates were excluded in the same order and the same candidates returned as at the actual election. The same results would, therefore, have been attained and much labour saved if the rules of the Municipal Representation Bill had been used. This committee, however, in view of the fact that the more exact method had already been established in Tasmania, and that the ascertainment of the results only involved an expenditure of a few hours more time, and that there were no data available to show the frequency of close contests in which a small change in the distribution of votes might possibly affect the result, recommended that no change should be made in the law. Still it would seem that the rules of the Municipal Representation Bill are sufficiently exact for all practical purposes except where the number of electors is small. The fractional transfer is of course the most perfect from the mathematical point of view, but the Royal Commission on Electoral Systems, after a careful examination of its working, report that "we agree with the Proportional Representation Society in regarding the additional labour involved as greater than it is worth."[10]
Where the number of electors is small, however, it is not only desirable to carry out the transfers with the exactness prescribed by the Tasmanian rules, but in important elections, such as those of the Senators in South Africa, it is desirable to introduce a further modification. In transferring the votes in ordinary elections fractions of votes are ignored, because such fractions do not affect the result. Where, however, there are only a few electors such fractions may become important, and, for this reason, the regulations (see Appendix IX.) adopted by the South African Government for the election of Senators provided that each ballot paper should be treated as of the value of 100, or, in other words, that fractions should be taken into account as far as two places of decimals. The application of these regulations presented no difficulty; the counting of the votes in each of the four Colonies proceeded without the slightest hitch.
The Gove or Dobbs Method.
The methods of transfer hitherto described all enable the voter to maintain complete power over the disposal his vote. It has, however, been suggested that the candidate for whom the vote is recorded should have the privilege of deciding to whom it should be transferred. The suggestion was first made by Mr. Archibald E. Dobbs, who, in 1872, in a pamphlet entitled General Representation, made the proposal that before the date of the election each candidate should publish a schedule of the names of any of the other candidates to whom he desired his vote to be transferred. This method of transfer by schedule is usually known as the "Gove" method, and was contained in the Bill submitted by Mr. W. H. Gove to the Legislature of Massachusetts, in 1891. Section 7 of this Bill reads as follows: "Votes shall be transferred according to the request of the candidate for whom they were originally cast to a person named in the list furnished by said candidate before the date of the election." With this method the elector in recording his vote for any one candidate would have no independent power of indicating to whom the vote should be transferred, and Mr. Dobbs, in a later pamphlet[11] has suggested that the elector should be given the option of accepting the schedule of preferences published by the candidate, or of indicating his own. Mr. Dobbs thus gets rid of the compulsory acceptance of a schedule of preferences, a proposal to which most English-speaking electors would have an instinctive dislike. But even to an optional schedule certain objections remain. The system has lost in simplicity, and the order of the candidates in the particular schedules would be determined in most cases by the party organizations.
The transferability of votes is the connecting link between all these systems; it is the essential feature upon which depends the proportionate representation of the contending parties, and the mode of transfer is properly regarded as a matter upon which different views may be held. As regards the second and third systems of transfer outlined above—which so far are the only ones which have been put into practice—experience confirms the theoretical conclusions of mathematicians that, save in the case of small electorates, both methods yield the same result. The second method was that used by the Proportional Representation Society for the purpose of its model elections, and is now applied in the election of Municipal Councils in Johannesburg and Pretoria. A description of the Model Election of 1908 will serve to illustrate the various processes involved in the sorting and counting of votes.
The model election of 1908.
In this election it was assumed that the voters in a constituency returning five members were asked to make their choice among twelve candidates. These candidates were all well-known political men, and were chosen with an attempt at impartiality from the Liberal, the Unionist, and the Independent Labour parties. As no Irish newspaper was publishing the ballot paper, no Nationalist was included.[12] This ballot paper, a copy of which appears on page 147, was sent, accompanied by a short explanatory article, for publication to, and appeared in, the following newspapers: The Times, The Morning Post, The Spectator, The Nation, The Daily News, The Financial News, The Manchester Guardian, The Yorkshire Post, The Yorkshire Daily Observer, The Western Morning News, The Western Daily Mercury, The Glasgow Herald, The Dundee Advertiser, The Woolwich Pioneer, and The Labour Leader. Readers of the newspapers were asked to cut out the ballot paper, mark it and return it to Caxton Hall by the first post on the morning of Tuesday, 1 December 1908. Ballot papers were also circulated independently among members of the Proportional Representation Society and their friends. About 18,000 papers were returned by newspaper readers, and about 3700 by members of the Society and their friends. In all a constituency of 21,690 electors was formed, a number whose votes were enough, but not too many, for counting in a single evening.
PROPORTIONAL REPRESENTATION ELECTION, 1908
BALLOT PAPER
PLEASE VOTE
In this Illustrative Election FIVE members are to be elected for a single constituency, such as Leeds. The following TWELVE Candidates are supposed to have been nominated.
Order of
Preference. Names of Candidates
……….. ASQUITH, The Rt. Hon. H. H.
……….. BALFOUR, The Rt. Hon. A. J.
……….. BURT, The Rt. Hon. Thomas
……….. CECIL, Lord Hugh
……….. HENDERSON, Arthur
……….. JONES, Leif
……….. JOYNSON-HICKS, W.
……….. LLOYD GEORGE, The Rt. Hon. D.
……….. LONG, The Rt. Hon. Walter H.
……….. MACDONALD, J. Ramsay
……….. SHACKLETON, David
……….. SMITH, F.E.
INSTRUCTIONS TO VOTERS
A. Each Elector has one vote, and one vote only.
B. The Elector votes
(a) By placing the figure 1 opposite the name of the candidate he likes best.
He is also invited to place
(b) The figure 2 opposite the name of his second choice.
(c) The figure 3 opposite the name of his third choice, and so on, numbering as many candidates as he pleases in the order of his preference.
N.B.—The vote will be spoilt if the figure 1 is placed opposite the name of more than one candidate.
* * * * *
This Ballot Paper should be filled in and returned not later than Tuesday, first post, 1 December 1908, in open envelope (halfpenny stamp), addressed to
THE RT. HON. LORD AVEBURY, Caxton Hall, Westminster, S.W.
The counting of the votes. General Arrangements.
The votes were counted at the Caxton Hall, Westminster, on the evening of Thursday, 3 December. Unfortunately, it was not found possible for all the newspapers to reproduce the ballot paper in its exact dimensions, and the unevenness in the sizes of the papers, which would not occur in a real election, caused some trouble to the counters. The method on which the room was arranged may best be gathered from the plan shown on next page.
[Illustration: ILLUSTRATIVE ELECTION, DECEMBER 3RD, 1908 PLAN OF ROOM]
In the centre of the room was the sorting table, where the votes were in imagination discharged from the ballot boxes. At this table were stationed a number of helpers, chiefly Post Office sorters, who through Mr. G. H. Stuart, of the Postmen's Federation, and Mr. A. Jones, of the Fawcett Association, had kindly volunteered their services. Here also were a dozen sets of pigeon-holes, each set having twelve compartments, and each compartment being labelled with the name of a candidate. As soon as the count began, the sorters started sorting the ballot papers according to the names marked 1, placing in each candidate's compartment the papers in which his name was so marked, and setting aside spoilt or doubtful papers. Printed instructions to the sorters had been issued, thus:—
1. Sort the ballot papers according to the names marked 1.
2. Place spoiled or doubtful papers on top of the case (right-hand side).
As the papers were sorted the two assistants supervising these processes took them to the small tables (checking and counting tables) ranged on either side of the sorting table. These tables were appropriated to the various candidates, and when it was expected that a candidate would poll a large number of votes—e.g., in the cases of Mr. Asquith and Mr. Balfour—several tables were allotted to him. At each of these tables sat two counters who acted in accordance with the following instructions:—
1. Count the papers into bundles of fifty.
2. See that the figure 1 appears against the name of the candidate whose papers are being counted.
3. Place mis-sorts at the side of the table.
4. Count each bundle twice.
5. Place on the top of each bundle a coloured slip bearing the candidate's name (already printed).
6. Note the final bundle with the number of papers therein contained.
The counters thus checked the accuracy of the sorters' work, and labelled the bundles of each candidate's votes with a card of a distinctive colour bearing his name. These bundles of votes were then taken to the returning officer's table, where there awaited them a row of twelve deep, three-sided open boxes, each labelled with the name of a candidate. The returning officer's assistants at this table made up the bundles of 50 into parcels of 500, and ascertained the total number of votes for each candidate, carefully keeping each candidate's papers in his own allotted box.
Lastly, the results as ascertained were shown on large blackboards. If and whenever any doubt arose as to the validity of a vote, it was taken to the returning officer by the supervisors and adjudicated upon by him. The accuracy of the sorting may be judged by the fact that when the 9043 votes attributed to Mr. Asquith on the first count were subsequently analyzed, it was found that only one paper was wrongly placed to his credit, a Liberal vote which should have gone first to Mr. Lloyd George.
As to these arrangements, one suggestion may be made for the guidance of future returning officers: it was found in practice that the work at the returning officer's table was too heavy for the two assistants to keep pace with the rapidity with which the votes were sorted and counted. Two assistants are required for the purpose of keeping a record of the various processes; two others for receiving and distributing the ballot papers.
The first count.
The first duty of the returning officer, as already explained, was to ascertain the total number of votes polled by each candidate, each ballot paper being a vote for the candidate marked 1 thereon. This was a simple task, which took about an hour and a quarter, and yielded the following result:—
Asquith (Liberal) 9,042
Balfour (Unionist) 4,478
Lloyd George (Liberal) 2,751
Macdonald (Labour) 2,124
Henderson (Labour) 1,038
Long (Unionist) 672
Hugh Cecil (Unionist Free Trader) 460
Shackleton (Labour) 398
Burt (Liberal) 260
Leif Jones (Liberal) 191
Smith (Unionist) 164
Joynson-Hicks (Unionist) 94
———
Total 21,672
The Quota.
It will be seen that, with this method of election, the general result, showing the relative strength of the parties, can be quickly ascertained, but, some time elapses before the definitive result, with the names of all the successful candidates, can be published. The first step necessary in determining which candidates were successful was to ascertain the quota, and this, in accordance with the rule above stated,[13] was found by dividing the total number of votes by six and adding one to the result. The number was found to be 3613, and the table given above shows that on the first count Mr. Asquith and Mr. Balfour had each polled more than a quota of votes. Both these candidates were, in accordance with the rules, declared elected, and, as some misapprehension prevails on this point, it should be stated that the order of seniority of members elected under this system would be determined by the order in which they were declared elected. In this case Mr. Asquith and Mr. Balfour would be the senior members in the order named.
The transfer of surplus votes.
The peculiar feature of the single transferable vote now came into play. Both Mr. Asquith and Mr. Balfour had polled more votes than were sufficient to ensure their election, and in order that these excess votes should not be wasted and a result produced such as that already shown to be possible where the votes are not transferable, it was the duty of the returning officer to transfer these surplus votes, and in doing so to carry out strictly the wishes of the electors as indicated on their ballot papers.
The largest surplus, that of Mr. Asquith, was first dealt with, and the transfer of votes, as already mentioned, was effected in accordance with the provisions of Lord Courtney's Municipal Representation Bill. All the votes recorded for Mr. Asquith were re-examined, all the ballot papers contained in his box being taken to the central table and re-sorted according to the next available preferences indicated by the electors. For this purpose the names of the elected candidates were removed from their former pigeon-holes, and one of the compartments vacated was marked "exhausted" and used as a receptacle for those papers which contained no available next preference. The instructions to sorters were:—
1. Sort the ballot papers according to the highest available preference.
2. When no further preference is indicated, place the ballot paper in the compartment marked "exhausted."
The term "next available preferences" needs definition. As a rule the next preference was the candidate marked with the figure 2; but if any supporter of Mr. Asquith had indicated Mr. Balfour (already elected) as his second choice, then the elector's third choice became the "next available preference." The papers for each next preference were made into bundles of 50, but, instead of a coloured card with the name of the candidate, a white "transfer" card was placed with each bundle. The transfer card was marked with the name of the candidate whose papers were being re-sorted and also with the name of the candidate who had been indicated as the next available preference. The instructions issued to the counters were as follows:—
_(a)_1. Check the sorting of the papers, i.e., see that the candidate whose papers are being counted is the highest available preference.
2. Place mis-sorts at the side of the table.
(b) 1. Count the papers into bundles of fifty.
2. Count each bundle twice.
3. Place on the top of each bundle a "transfer card" showing from and to whom the votes are being transferred.
4. Note each bundle with the number of papers therein contained.
These bundles were placed in a second series of open boxes on the returning officer's table, each box being labelled with the name of a candidate and being smaller in size than the boxes containing the first preferences. The number of next available preferences for each candidate was then ascertained. It was, of course, not the duty of the returning officer to transfer all the re-sorted papers; it was necessary to retain a "quota" for Mr. Asquith; and an operation which requires some care now took place. The papers contained in each of the second series of boxes were divided into two portions, bearing in each case the same proportion to one another. One portion was transferred to the candidate who had been indicated as the next preference, and the other was placed in Mr. Asquith's box, the portions reserved for him constituting his quota; the actual papers transferred to each next preference were those last placed in the box bearing his name. The details of this process are set forth in the table overleaf.
PROPORTIONAL REPRESENTATION ELECTION, 1908
TRANSFER SHEET
Distribution of the Rt. Hon. H. H. ASQUITH's surplus.
Surplus Votes 5429
No. of Papers showing a next preference 9009
Surplus 5429
Proportion to be transferred = ————————————- = ——
Total of next preferences 9009
Column Headings:
A. Names of Candidates indicated as next preference.
I. No. of papers on which Candidate is marked as next preference.
II. No. of Votes transferred to next preference. (Fractions ignored.)
III. No. of Votes retained for Mr. Asquith's Quota.
A. I. II. III.
Balfour, The Rt. Hon. A. J. — — —
Burt, The Rt, Hon. Thomas 468 282 186
Cecil, Lord Hugh 132 79 53
Henderson, Arthur 261 157 104
Jones, Leif 176 106 70
Joynson-Hicks, W. 17 10 7
Lloyd George, The Rt. Hon. D. 7,807 4,704 3,103
Long, The Rt. Hon. Walter H. 46 27 19
Madonald, J. Ramsay 51 30 21
Shackleton, David 35 21 14
Smith, F. B. 16 9 7
——- ——- ——-
Total of next preferences 9,009 5,425 3,584
Preferences exhausted . . 33 — 33 ——- ——- ——- Total 9,042 5,425 3,617[14]
This table needs, perhaps, a further word of explanation. The first column shows the result of the re-sorting of Mr. Asquith's papers, Mr. Burt having been indicated as the next preference on 468 papers, Lord Hugh Cecil on 132 papers, and so on. The papers for each next preference were, as already staked, divided into two portions, and the second and third columns show the result of this division. The division is carried out in a strictly proportional manner, according to the following principle. If 5429 surplus votes are to be transferred from a total of 9009 unexhausted voting papers, what portion should be transferred from 468, from 132, and so on. The proper numbers, which are given in the second column, are found by a simple rule of three process; each of the numbers in the second column is obtained from the corresponding number in the first column by multiplying by the fraction 5429/9009, that being the fraction which represents the proportion of unexhausted papers to be transferred. The figures in column III., which are the votes retained in each case to make up Mr. Asquith's quota, are obtained by subtracting the corresponding numbers in column II. from those in column I. Ten separate calculations were thus necessary, and for this part of the election it is desirable that the returning officer should have two assistants who are accustomed to figures. These should check one another's work. In Belgium the returning officer is assisted by two "professional calculators."
The ballot papers with the votes constituting Mr. Asquith's quota were replaced in his original box and never touched again. The ballot papers transferred were placed in each case on the top of the papers already contained in the box of the candidate to whom the transfer was made.
As the result of the transfer of Mr. Asquith's surplus it was found that the total of Mr. Lloyd George's votes amounted to 7455, and as this number exceeded the quota, Mr. Lloyd George was declared elected, he being the third member chosen. Mr. Balfour's surplus was then distributed in a similar manner. The number of votes transferred is shown in the result sheet, pp. 160-61. As Mr. Lloyd George's total exceeded the quota, it was also necessary to dispose of his surplus. In the latter case only the papers transferred to Mr. Lloyd George, and not his original votes, were re-examined, as his surplus consisted of votes originally given to Mr. Asquith.
The poll now stood:—
Asquith (Liberal) 3,613 \
Balfour (Unionist) 3,613 > Elected
Lloyd George (Liberal) 3,613 /
Macdonald (Labour) 2,387
Henderson (Labour) 2,032
Burt (Liberal) 1,793
L. Jones (Liberal) 1,396
Long (Unionist) 1,282
Cecil (Unionist Free Trade) 822
Shackleton (Labour) 683
Smith (Unionist) 258
Joynson-Hicks (Unionist) 167
Votes lost through neglect of fractions 13
It will readily be seen that these transfers have been in accordance with what might have been assumed to be the general political preferences of the electors. The Liberal surplus votes from Mr. Asquith naturally went on chiefly to Mr. Lloyd George, and the overflow from Mr. Lloyd George, after filling up his quota, went on to Mr. Burt and Mr. Leif Jones, whose positions were greatly improved in consequence, though neither obtained the quota. At the same time a formidable addition of 834 votes was given to Mr. Henderson, the votes doubtless of Liberal sympathisers with Labour; and Lord Hugh Cecil received 88 votes, presumably from moderate Liberals who lay chief stress on Free Trade. On the other hand, Mr. Balfour's smaller Unionist surplus was divided mainly between Mr. Walter Long, who received 526 additional votes, and Lord Hugh Cecil, who received 195.
The elimination of unsuccessful candidates.]
After the transfer of all surplus votes had been completed, the work of the returning officer again became very simple. Three members only had been elected, two more were required, and there remained in the running nine candidates, none of whom obtained a quota of votes. Another process now began, namely the elimination of candidates at the bottom of the poll, beginning with the lowest and working upwards. The group of electors who have recorded their votes for the candidate lowest on the poll are evidently not sufficiently numerous to have a direct representative of their own. The process of elimination allows these electors to re-combine with other groups until they become part of a body large enough to be so entitled. The supporters of the lowest candidate are treated as being asked (and answering, if they care to do so, by their next preferences) the question: "The candidate of your first choice having no chance of election, to whom now of the candidates still in the running do you prefer your vote to go?" By this process, first the two candidates, Mr. Smith and Mr. Joynson-Hicks, who at this stage were at the bottom of the poll and whose combined votes were less than those of the third lowest candidate, were eliminated and their votes transferred to the next preferences of their supporters. No one was elected as a result of this operation, and accordingly the votes of Mr. Shackleton and Lord Hugh Cecil, now lowest on the poll, were transferred in the order named.
These and all other eliminations were of the same character. All the papers of the eliminated candidates which showed an available next preference were transferred, and no calculations such as were required in the case of the transfer of surplus votes were needed. It will be sufficient if the details of one process—the transfer of Mr. Shackleton's votes—are given; for the details of all other similar transfers the full table on pp. 160-61 should be consulted. The votes of Mr. Shackleton were disposed of as follows:—
TRANSFER OF MR. SHACKLETON'S VOTES
Names of Candidates Number of Papers indicated as next for each next preference. preference.
Burt 89
Cecil 18
Henderson 233
Jones 57
Long 8
Macdonald 252
Preferences exhausted 45 —- Total 702
The transfers of the votes both of Mr. Shackleton and of Lord Hugh Cecil were completed, but still no fresh candidate had the quota, and Mr. Lief Jones's 1500 votes came next for distribution. These 1500 votes might have been expected to go to Mr. Burt, the sole remaining unelected Liberal, who had already 2025 votes, and make his election practically secure. But here came a surprise; Mr. Leif Jones's supporters (who had, of course, in most instances, come to him from Mr. Asquith and Mr. Lloyd George) had in some cases marked no further preferences, so that their votes were no longer transferable, and in many other cases had marked Mr. Henderson or Mr. Macdonald as their next preference; thus at the conclusion of this operation the result of the election was still doubtful.
Two places had still to be filled, and the poll stood:—
Asquith (Liberal) 3,613 \
Balfour (Unionist) 3,613 > Elected
Lloyd George (Liberal) 3,613 /
Macdonald (Labour) 2,851
Henderson (Labour) 2,829
Burt (Liberal) 2,683
Long (Unionist) 2,035
Mr. Long's votes had now to be distributed; the majority of his supporters were Unionists who had not marked any preference for either of the two remaining Labour candidates or for the remaining Liberal candidate, and their votes consequently were not capable of being transferred. But some 370 of Mr. Long's supporters had shown a preference for Mr. Burt (presumably as being reckoned not so Socialistic as his competitors) as against some 27 for Mr. Macdonald and 80 for Mr. Henderson, so that the poll stood:—
Asquith (Liberal) 3,613 \
Balfour (Unionist) 3,613 > Elected
Lloyd George (Liberal) 3,613 /
Burt (Liberal) 3,053
Macdonald (Labour) 2,938
Henderson (Labour) 2,910
Mr. Henderson, being at the bottom of the poll, was then eliminated, but it was unnecessary to proceed with the transfer of his votes as, after his elimination, there were only five candidates remaining, and five was the number of members to be elected. The work of the returning officer was at an end, the following candidates being elected:—
Asquith (Liberal)
Bafour (Unionist)
Lloyd George (Liberal)
Burt (Liberal)
Macdonald (Labour)
The whole process of the election is shown by the returning officers' full result sheet.
The fairness of the result.
The fairness of this method of voting is at once apparent. Each group of electors as large as a quota secured a representative. The Liberals were in a very large majority, and with the block system and probably with the single-member system would have nominated five candidates and have obtained all five seats. In this election the two smaller groups, the Unionist and Labour parties, each returned one member. The voters did not, in recording their preferences, restrict themselves to candidates of one party, but nevertheless, it will be of interest to compare the seats gained with the strength of parties as indicated by the first preferences. The party vote disclosed in the first count was as follows:—
Votes polled.
Liberal 12,244
Unionist 6,868
Labour 3,660
———
Total 21,672
The quota was 3613, and these totals show that the
Liberals obtained 3 quotas with 1405 votes over and gained 3 seats.
Unionists obtained 1 quota with 2265 votes over and gained 1 seat.
Labour obtained 1 quota less 53 votes and gained 1 seat.
PROPORTIONAL REPRESENTATION ELECTION, 1908—RESULT SHEET
No. of Votes,—21,672.
No. of Seats—5.
Quota = (21,672/6) + 1 = 3613
Col 1: First Count
Col 2: Transfer of surplus votes (Asquith's)
Col 3: Result
Col 4: Transfer of Surplus Votes (Bafour)
Col 5: Result
Col 6: Transfer of Surplus Votes (Lloyd George)
Col 7: Result
Names of Candidates. 1 2 3 4 5 6 7
Asquith, The Rt.Hon.H.H. 9,042-5,429 3,613 — 3,613 — 3,613
Balfour, The Rt.Hon.A.J. 4,478 — 4,478-865 3,613 — 3,613
Burl, The Rt. Hon. Thomas. 260 +282 542 +12 554+1,239 1,793
Cecil, Lord Hugh 400 +79 539+195 734 +88 822
Henderson, Arthur 1,038 +157 1,195 +3 1,198 +834 2,032
Jone, Leif 191 +157 297 +2 299+1,097 1,396
Joynson-Hicks, W. 94 +10 104 +52 156 +11 167
Lloyd George, The Rt.Hon.D. 2,751+4,704 7,455 — 7,455-3,842 3,613
Long, The Rt.Hon. Walter H. 672 +27 699+520 1,225 +57 1,282
Macdonald, J. Ramsay 2,124 +30 2,154 +5 2,159 +228 2,387
Shackleton, David 398 +21 419 +2 421 +202 683
Smith, F.E. 184 +9 173 +65 238 +20 258
Votes lost through neglect of fractions - +4 4 +3 7 +6 13
Preferences Exhausted - - - - — — —
Totals 21,072 - 21,672 — 21,672 — 21,672
Col 8: Transfer of votes (J Hicks and Smiths)
Col 9: Result
Col 10: Transfer of Votes Shackleston's)
Col 11: Result
Col 12: Transfer of Votes (cecil's)
Col 13: Result
Col 14: Transfer of Votes (L.Jones)
Col 15: Results
Col 16: Transfer of Votes (Long's)
Col 17: Final Result.
8. 9. 10. 11. 12. 13. 14. 15. 16. 17.
Asquith — 3,613 — 3,613 — 3,613 — 3,613 — 3,613 E
Balfour — 3,013 — 3,613 — 3,613 — 3,613 — 3,613 E
Burl. +21 1,814 +89 1,903+122 2,025 +658 2,683 +370 3,053 E
Cecil +88 908 +18 923-926 — — — — —
Henderson +14 2,046+233 2,270 +49 2,328 +501 2,829 +81 2,910
Jone +12 1,408 +57 1,465 +35 1,500-1,500 — — —
Joynson-Hicks 167 — — — — — — — — —
Lloyd George — 3,613 — 3,613 — 3,613 — 3,613 — 3,613 E
Long +233 1,505 +8 1,513+490 2,003 +32 2,035-2,035 —
Macdonald +21 2,408+252 2,680 +48 2,708 +143 2,851 +87 2,938 E
Shackleton +19 702-702 — — — — — — —
Smith -258 — — — — — — — — —
Votes lost — 13 — 13 — 13 — 13 — 13
Exhausted +29 29 +45 74+182 256 +166 422+1,497 1,919
Totals — 21,672 — 21,672 — 21,672 — 21,672 —21,672
This result is as fair as is possible, and would have been equally attained if, as would probably be the case in a real election, there had been but little cross voting. The total results in the Tasmanian General Election, 1909 (six-member constituencies) showed an exact proportion between the votes polled and the seats gained by the respective parties.[15]
Improved arrangements in the Transvaal elections.
The arrangements made at the model election were adopted by the Chief Electoral Officer of Tasmania,[16] and were also adopted by the returning officers of Pretoria and Johannesburg. Experience has shown that some improvements in details can be made. Both at Pretoria and Johannesburg less work was done at the returning officer's table. The counters were placed more directly arrangements under the superintendence of the returning officer's assistants, and the final totals of each operation were ascertained at the counters' tables. When the ballot boxes were brought in by the presiding officers of the polling stations with a return of the votes they contained, the returning officer handed them one by one to superintendents who took them to that section of the counting force over which they had charge. The counters ascertained the number of papers in each ballot box. The superintendents reported the total number to the returning officer, and if this number agreed with the presiding officer's return the ballot box and contents were handed back to the returning officer. After the contents of all the ballot boxes had been verified and the grand total of votes ascertained, all the papers were emptied into one box and were well mixed. The papers were then sorted at a central table, as in the election already described; the superintendent took the papers to the counters, each of whom ascertained the number of votes for that candidate whose papers he had been deputed to count. The superintendents brought a statement of the totals for each candidate to the returning officer, and if the aggregate of these figures did not agree with the number of ballot papers distributed to the sorters a fresh count was ordered. The elections at Johannesburg and Pretoria demonstrated that the requisite accuracy in counting could be easily attained. The operations were characterized with remarkable precision. There was no error in the counting of the votes at Pretoria during the whole of the operations, and the same remark holds good of Johannesburg, save that one ballot paper which had been accidentally torn was omitted to be counted. The two pieces had been pinned together, and the paper, which in consequence had been rendered shorter than the others, was overlooked. The omission was quickly discovered, and no other error took place during the whole of the proceedings. The various counting processes check one another. Any errors occurring in the earlier operations are thrown out in the course of the subsequent proceedings, for the totals of the votes at the conclusion of each operation must agree with the total shown at the commencement of the count. In another feature the organization of the Transvaal elections might be copied. All spoilt or doubtful papers were brought to the returning officer's table by his assistants, and were not examined until the conclusion of the first count. The whole of these papers were then gone through by the returning officer, who decided the question of their validity in the presence of the candidates or their representatives. The returning officer also examined all papers which were treated as "exhausted," but this work might have been deputed to the assistant returning officer.[17]
Criticisms of the single transferable vote.
After reviewing the whole of the evidence submitted to them, the Royal Commission on Electoral Systems reported that "of schemes for producing proportional representation we think that the transferable vote will have the best chance of ultimate acceptance," but the Report contains some criticisms of its mechanism which demand consideration. These criticisms are directed to two points: (1) the effect of later preferences in deciding the result of an election; (2) the process of eliminating candidates at the bottom of the poll.
Effect of late preferences.
The Royal Commission express the opinion that late preferences may have an undue weight in deciding the result of an election. But the Commissioners seem to have been unnecessarily alarmed in this matter. A careful analysis of the preferences recorded in the Tasmanian elections was made by a Committee appointed for the purpose by the Tasmanian Government. This Committee ascertained that the comparative values of the various preferences in determining the result of the election were as follows:—
1st preference .739 2nd .140 3rd .051 4th .029 5th .014 6th .008 7th .009 8th .008 9th .003
In other words 73.9 per cent, first preferences became effective votes, 14.0 per cent, second preferences became effective votes, and so on. These figures show the great superiority in value of the earlier preferences, and this superiority was also seen in the Transvaal elections. In Pretoria 68 per cent, of the first preferences were directly effective in returning candidates, in Johannesburg 67.5 per cent. Second preferences primarily come into play in favour of candidates of similar complexion to the candidates first chosen, and when, as is possible in the last resort, a vote is passed on in support of a candidate of a different party, this is no more than the Commissioners themselves approve and recommend for adoption in the case of three or more candidates standing for a single seat. The difference between the effect of the final transfers under a system of proportional representation and of transfers under the system recommended by the Commission is that in the first case they might determine the character of one out of five or more members representing a constituency, in the other they might affect the representation of each of the five or more divisions into which the constituency would be divided.
The elimination of candidates from the bottom of the poll.
The second criticism concerns the elimination of candidates. It is sometimes contended that it is unfair to eliminate the candidate at the bottom of the poll, because had he remained longer in the contest he might have received at the next stage a considerable amount of support. Taking an extreme case, the candidate at the bottom of the poll may have been so generally popular as to have been the second choice of the majority of the electors. This is theoretically conceivable, but it does not conform to the facts of elections. The principle of eliminating a candidate at the bottom of the poll is not peculiar to the single transferable vote. When a constituency returns but one member and there are three candidates, and it is desired by means of the second ballot to ensure the election of the candidate who commands the support of the majority of the electors, the candidate lowest on the poll is eliminated and a second ballot is held to decide between the claims of the remaining two candidates. In such a case it is conceivable that the candidate lowest on the poll may have been more acceptable to the majority of the electors than the candidate finally selected. But the system of the single transferable vote with constituencies returning several members diminishes very considerably any such possibility. In the first place, the candidate to be successful need only obtain a much smaller proportion of the total number of votes than in a single-member constituency. In the latter he must poll just over one-half before he is safe from defeat; in a seven-member constituency if he polls one-eighth he will escape this fate. The candidate who has a reasonable proportion of support, therefore, stands less chance of being excluded. In the second place no candidate is excluded until after the transfer of all surplus votes has been completed. If, in a constituency returning several members, a candidate, after the transfer of all surplus votes, is still at the bottom of the poll, the facts would seem to indicate that he was not even the second favourite of any considerable number of electors. The preferences actually given in elections show how little force this criticism possesses. The table below was prepared by the Committee appointed by the Tasmanian Government. It shows the result of an examination of all the votes cast in the district of Wilmot for the election of five members of the Tasmanian House of Assembly in April 1909. The names of the candidates are given with the numbers of the various preferences recorded for each candidate. The total number of second preferences recorded for Waterworth, the first candidate to be excluded, was 141. Similar tables for the other four districts show that no injustice arose from the exclusion of the lowest candidate. The only occasion on which the criticism has any force is when, in filling the last seats, the conditions are analogous to those which obtain in a three-cornered fight in a single-member constituency. Yet in the latter case the Royal Commission did not hesitate to recommend the exclusion of the lowest candidate.
DISTRICT OF WILMOT: NUMBERS OF VARIOUS PREFERENCES
Name. Preferences.
1 2 3 4 5 6 7 8 9 10
Best 935 690 596 609 615 550 23 2 7 5
Dumbleton 518 537 603 632 819 650 24 4 3 5
Field 930 699 692 619 555 585 21 9 4 5
Hope 1,232 1,302 1,077 551 229 159 13 6 2 5
Jensen 1,955 894 1,087 132 58 58 13 19 7 36
Kean 599 1,521 1,370 118 53 50 11 28 38 15
Lee 822 750 902 618 512 488 27 4 7 1
Lyons 1,079 1,444 1,329 93 76 65 21 29 32 12
Murray 572 885 972 848 625 395 14 6 7 1
Waterworth 221 141 236 590 198 254 141 21 6 9
——- ——- ——- ——- ——- ——- —- —- —- —
8,863 8,863 8,863 4,810 3,740 3,254 308 128 113 94
The elimination of candidates has been criticized from another point of view. The Royal Commission, while careful not to endorse this criticism, and referring to it with reluctance, "because doubts about the absolute reliability of the mechanism of the system may arouse prejudices disproportionate to the importance of the subject, which is very small in comparison with the other considerations involved," review the evidence which had been submitted to them as follows: "The element of chance involved in the order of elimination is exceedingly difficult to determine. It would appear that the element is perceptible in certain contingencies, but the rarity or frequency with which these would occur in actual practice is a matter of pure speculation, as it apparently depends on the amount of cross-voting which voters permit themselves in the use of their later preferences, a point only to be decided by experience. 'Chance' in this connexion has not quite the same meaning as when used in respect of the method of transfer. In the case of the latter we were dealing with mathematical probabilities; the chance which is said to be involved in the process of elimination consists in the fact that the results of the election may vary according to the strength of quite irrelevant factors. Thus, a case was put to us to show that with certain dispositions on the part of the electors the representation of a party might be so much at the mercy of the order of elimination that while it would only obtain one seat with 19,000 votes of its own it would obtain two with 18,000, because in the latter case the order of elimination of two candidates would be reversed."[18]
It is here suggested that the results may depend upon the amount of cross-voting which voters would permit themselves in the use of their later preferences. The whole paragraph abounds in obscurities, and the word "cross-voting" is used in such a context as to make it quite uncertain whether the Commission mean by it inter- or intra-party voting, or both. It is somewhat difficult to make a definite answer to a charge so indistinctly formulated. Cross-voting, in the ordinary sense, may certainly affect the result. If the supporters of a Radical candidate prefer to give their second preferences to a Labour candidate rather than to a moderate Liberal, such cross-voting obviously may determine whether the Labour candidate or the moderate Liberal will be successful. There is no element of chance involved. The object of the system is the true representation of the electors, and the returning officer must give effect to their wishes. The numerical case cited by the Commissioners can only occur when so-called supporters of the party in question are so indifferent to its fate as to refrain from recording any preferences for any members of the party other than their own favoured candidate. Such voters can hardly be called "members of a party" for the purpose of contrasting its strength with that of another party.[19] Even such cases, supposing them at all probable in practice, could be provided against, as has been suggested by Mr. Rooke Corbett of the Manchester Statistical Society, by determining a new quota whenever any votes have to be set aside as exhausted. But the elections in which the system has been tried show how little these cases accord with the facts. The large number of exhausted papers which occur in the model election described in this chapter, which was organized through the press, perhaps accounts for much of this criticism. In real elections the percentage of exhausted papers is much less. Thus in Johannesburg, where one rigidly organized party, another party more loosely organized, and ten independent candidates took the field, the electors made good use of their privilege of marking preferences. Some 11,788 votes were polled. At the conclusion of the tenth transfer only 104 votes had been treated as exhausted. In Pretoria, where there were 2814 votes, the total number of exhausted votes at the end of the election was only 63. This happened on the occasion of the first trial of the system in Johannesburg and Pretoria, and further experience will lead to an even fuller exercise of the privilege of marking preferences. There is no case for a criticism based on such a hypothetical example as that hinted at by the Commission.