The average, it appears, is by no means a typical one, except in the case of the counties, where the large number immediately above it neutralises to some extent the numbers below. The balance is here affected also by the two large Essex constituencies bordering on the metropolis. Conversely, in the boroughs, the 12 small seats at the bottom of the scale are balanced by the 29 at the other end. The metropolitan figures show a very material discrepancy in the opposite direction, which is to some extent due to the semi-suburban constituencies of Croydon, Wandsworth, Deptford, and the southern portion of West Ham. The agreement of the general borough average with that of the counties is therefore accidental. Thrown into a diagram of the usual construction, with the percentages vertically at right angles to the population-groups, the table presents a remarkable resemblance to the Zermatt view of the Matterhorn.

12. I may now pass on to the territorial division of the table of which the totals have been discussed above. This is based on the grouping adopted in the census and other general returns, with two comparatively trifling modifications. In the first place, Croydon and West Ham are included with the metropolis, in a separate group. In the second, the counties of Derby and Nottingham have been detached from the North Midland group in the usual arrangement, and that title is assigned to them alone, the rest of the group being termed the East Midlands. For the former innovation the reasons are obvious, as Croydon and West Ham have long been practically joined to London in all but administrative arrangements. For the second departure from custom I am indebted to recess oratory, not merely of this or last year, but for some time past, when authoritative and responsible statesmen have impressed upon their adherents the importance of remembering that England is divided politically by the river Trent. Until this distinction was so prominently brought to notice, I thought that that "smug and silver" river rained its influence, consisting, I believe, of beer and grayling, impartially on both banks. I must admit, however, that I am wrong. Either the ale is potent on one bank only, or the fish, though as Canning observed, they drink much and say little, have ceased to be a political feature. At all events, the statistics show that the difference between the two tracts is quite a recognisable one, and worth maintaining in the tables I am now submitting, though not perhaps to the extent or in the full sense in which it was understood by those whose precepts I am following in the matter. The following statement, which 1 will not stop to discuss, sufficiently indicates the results of the growth of the great manufacturing centres in the north. It should be compared with

the corresponding figures for the whole country already given in Table II.

TABLE VI.-Growth of Population &c., North of Trent.

13. Maintaining the distinctions of county, borough and metropolitan constituencies, the population, voting strength and representation of the cleven groups are summarised in Table VII.

In this table we are brought face to face with one of the main difficulties which are met with in any attempt to deal satisfactorily with the statistics of representation, namely, the weight to be assigned to the vote of the two-member constituencies, to which I must refer for a moment. Of these boroughs there are twenty-one, including the City of London. The usual method of dealing with them is to ignore the second seat, and to count the electors only. This slightly disturbs the relation between population and the franchise, still more that between electors and votes. The size of at least half the towns here included would give them, if other considerations were equal, two, and in some cases, three, seats, all above the average in point of population and electors. Others, like Derby and Southampton, are approaching this limit and, at a little distance below again, come Plymouth and Halifax. On the other side of the balance we have Ipswich and Bath, neither of which come up to the average of the single constituency on the general scale, and York, Devonport and Northampton, which do not far exceed it. On the assumption, then, that the subdivision of these larger towns into two constituencies for the purpose of compilation will, in their aggregate, counterbalance the twelve minute constituencies of less than 20,000 inhabitants, I have in every case taken the voting strength of the whole town to be double the number of individual electors, and for this reason bave headed the columns in question not "electorate" but "voting strength." I shall have to enter into further explanations as to the distribution and adjustment of the votes polled on the two occasions with which I have to deal, when I reach that portion of my paper.

VOL. LIX. PART I.

2

14. Reverting to Table VII above, the best way of comparing the relative position of the different groups in respect to the three main features presented in it, is to show the figures in their proportional form, as in the following table:

TABLE VIII.—Distribution per Cent. of Population, &c.

Population.

Voting Strength.

Representation.

Group.

County. Borough. Total. County. Borough. Total. County. Borough. Total.

Amid the general agreement of the three ratios with each other in the various divisions, a few cases of discrepancy stand out from the rest. For instance, London, judging by population only, is deficiently represented. On the other hand, taking as the test the roting strength, it has more than its share. In the country north of the Trent, the voting strength is above and the representation below the population ratio, especially as regards the boroughs. If we take the groups separately, it will be seen that the main disproportion is found in Yorkshire and the Lancashire division. In both, the boroughs agree fairly well in respect to population and voting strength, but, whereas Yorkshire has a representation figure of 8.5 against the voting figure of 5.8, Lancashire and Cheshire show only 57 for representation, against over 9 for their voting strength. The latter group falls short too in its county representation, but Yorkshire in its voting strength only. In the West Midland and the South Western groups there is an apparent irregularity, mainly traceable in the borough constituencies, and due apparently to the presence of such small seats as Shrewsbury, Hereford, Kidderminster, Stafford, and Warwick in the one, and Salisbury, Taunton, Bath, and Falmouth in the other. A similar remark is applicable, so far as I can see, to the South Midland group, certainly Bedford, Peterborough, and the two seats of

Northampton. But the most noteworthy proportions are those I have mentioned above-London, Lancashire, and Yorkshire.

15. The ratio of voting strength to population in each group is not a feature which presents great variation, so I have not burdened the record with a special table regarding it. I may cursorily mention, however, that amongst counties, Yorkshire, the West Midlands, and, above all, the East Midland groups, show the highest proportions. In the last case it reaches 21 per cent. as compared with the general mean of 184. The ratio is remarkably low in the Northern group, where it is but 14 per cent., only reaching 164 in the Lancashire group. Amongst the borough groups London, of course, stands out pre-eminent in paucity of electors, with only 13.6 per cent. The West Midlands show only 156, and Yorkshire and Lancashire between 17 and 175. The Eastern counties, owing perhaps to two double constituencies in a small population, containing also such seats as Bury St. Edmunds, Lynn, and Colchester, has an average of 25 per cent. Leicester, again, and the North stand high, with over 22 per cent. each. On the whole, the irregularity in this ratio in the towns, as compared with the county constituencies, is not more than is to be expected, considering the diversities of conditions not only political but industrial also.

II. THE ELECTIONS OF 1892 AND 1895.

(a) General Considerations.

16. I have now set forth all that appears to me to be needed to link my description of the present conditions of the franchise and distribution question to the description of its. past conditions by my eminent predecessors who have dealt with that subject. I propose accordingly to henceforth consider the machine at work. It would be obviously futile to attempt to do this by analysing the results of a single election. But two elections in succession seem to form an adequate basis of comparison and average. If we look back over the contests of the last thirty years, we find that very often, one may almost say invariably, unless under conditions of special or temporary moment, one election is the complement of its predecessor. I do not mean, of course, as regards the work done, and that the process of "stealing clothes" or brooms, or what not, is the inseparable accompaniment of a change of government; I refer only to the mind of the average "wobbler," and he it is who decides elections. His opinions are, as a rule, limited by two trials, or, in other words, the mean of two elections will probably eliminate or smooth down the fluctuations of this all-important harmonic. After a period of rest, or when some wide enfranchisement has passed into