We can think of only one objection likely to be raised against the foregoing statistical argument. Some one may say that we have made too little use of the decade 1890-1900, but have preferred the score of years 1880-1900. It is true that the last decade (1890-1900) shows better for the Negro than the preceding (1880-90)--which, indeed, indicated his over-rapid decadence throughout the South. But it seems hardly possible that this showing should not be deceptive. For there is not a single known circumstance that favoured him in his last decade rather than in the preceding. The explanation seems very simple; the coloured returns of the eleventh census were incomplete--not nearly so incomplete as those of the ninth, yet enough so in comparison with the tenth and the twelfth to make the showing for 1880 to 1890 too bad, and for 1890 to 1900 too good. The census reports of the Black population for 1850 and 1860 seem to have been substantially correct; for 1870, extremely incomplete; for 1880, greatly better; for 1890, not nearly so good; for 1900, much better again. For 1870 this is now conceded. Thus, in Mississippi the coloured population increased from 1850 to 1860 by 126,000; from 1870 to 1880 by 206,090; but from 1860 to 1870 by only 6,797 (impossible!). In Kentucky it actually lost (1860-70) 13,957, but gained (1870-80) 49,241; and once more lost (1880-90) 3,380. So, in South Carolina, the Negro gain was (1870-80) nearly 190,000, but (1860-70) only 3,500. So, in Missouri, a gain of 28,463 (1850-60) and of 27,279 (1870-80), but (1860-70) a loss of 432. The indications of imperfection in the census of 1890 seem clear, though not so glaring as in that of 1870. Such, for instance, are the actual decreases in the Negro population of Kentucky and Missouri, and the extremely small gains (1880-90) of 5,500 (Maryland), 5,000 (Missouri), 4,000 (Virginia), against gains (1890-1900) of 19,500 (Maryland), 11,000 (Missouri), 25,000 (Virginia). Other imperfections, not so glaring, but quite as unmistakable, a careful eye may detect only too frequently. Thus, consider the following returns per 1,000,000 for the census years-- 1860 1870 1880 1890 Insane 765 971 1,833 1,697 Feeble-minded 602 636 1,533 1,526 Deaf and Dumb 408 420 675 659 Blind 403 527 976 805 In all these classes a steady increase up to 1880, then a sudden falling off in 1890. Once more, the death rate in the non-registration area in 1880 was 13.42 per thousand; in 1880 it was only 10.79. Such an improvement in health, especially in districts mainly rural, is quite incredible. The fact is that for many purposes of comparison the eleventh census is unavailable--a fact that greatly strengthens many of our contentions.