rim of lead 1 inch wide and one-sixteenth of an inch thick, soldered to the bottom of the cylinder on the inside. Two wires cross at the center of figure where the connecting line is attached. The construction of the upper float is shown by the drawing. The lower part is a semi-ellipsoid, the upper part a very flat cone, having an altitude of one-eighth of an inch on a base of 8 inches. A tube three-eighths of an inch in diameter passes through the center. The connecting wire was of annealed iron, No. 26, B. W. G., or .018 inch diameter, and the reel was used to avoid weakening the wire by kinking, as well as to facilitate handling. An average of the floats sent into the field gives the following: Lower float: Weight in air. Upper float: Weight in air.......... Additional weight required to just sink the combination..... Ounces. 38 12 26 9 24 3 The moment of stability of the lower float is about nine-elevenths that of the Ellis float, while it weighs considerably less in air and 50 per cent. more in water. It also presents an edge of one-sixteenth of an inch in width to the action of upward currents, and the air chamber is of a form which will resist the collapsing pressure of deep water. The upper float was designed to offer a minimum surface to the wind, and also to have its free buoyancy brought into play with the least possible vertical motion. In the field the floats were found to have one defect, that of great delicacy and corresponding difficulty in handling. Constant complaint was made of the fragility of the wire and the want of free buoyancy of the combination. Several sub-floats were lost from the former, and many observations partially or wholly lost from the latter cause. METHODS. In determining the sediment from the samples of water collected, the method followed was that of filtration. Experiments were made on several kinds of filters and various ways of handling them. The Swedish, two grades of German, and French filters were tried, and it was found that they were effective in the order named, but only the best Swedish gave a clear filtrate. Water passed through four French papers still had a perceptible quantity of sediment in it, while one Swedish paper was sufficient to remove every trace, except under the microscope, which showed a few grains of less than one twenty-thousandth of an inch in diameter. Numerous experiments upon drying, spontaneously and by absorbents, resulted unsatisfactorily, while showing that, for small quantities of sediment, the error introduced by an undetected change in the hygrometric state of the filter might seriously affect and even reverse the result. Recourse was then had to artificial heat, and the trials were so satisfactory that it was adopted. The filters were prepared by carefully counterpoising two of them by trimming and then comparing the rest with one of these standards and noting the difference of weight. The other standard was kept with the papers during the weighing and put into the scale pan from time to time, and if any difference of weight appeared it was noted and used as a correction. In treating the samples, the bottles were uncorked and the clear water carefully poured off into a measuring tube, graduated to cubic centimeters, and the volume noted. It was then poured back into the bottle and the tube rinsed out with filtered water, which was also poured into the bottle. The corks were then replaced and the bottles violently shaken, after which the contents were poured upon the filters, the bottles rinsed with filtered water, and the product added to the proper filters. The standards were put into the funnels and treated with the same amount of filtered water. While the papers were still wet they were transferred to the drying box, which had been previously heated to the proper temperature, when one of the standards was placed in the hottest and the other in the coolest part of the box. The two standards were taken out from time to time and placed in the scales. At first the one in the hottest position would weigh the less, but their weight gradually approached each other until they finally balanced. After this it was found that their balance remained perfect, no matter how far the drying was pushed. They were left in the box a few minutes after they had balanced, in order to insure the same hygrometric state in all the papers. When taken from the drying box the papers were placed on a rack beside the weighing case, one of the standards being placed in the center of the group and the other in the scale-pan. As soon as the papers were cool the standard on the rack was counterbalanced against the one in the case and the difference noted. The papers were then compared with the standard in the case and the difference of weight noted. The comparison of the standard was repeated two or three times during the weighing and at the end. The results gave a column of corrections which, when applied to the observed weights, transferred them from the standard in the case to that in the rack. By this indirect process the papers were compared with a standard which remained in the rack with them. One source of error was detected and guarded against. When the papers contained a considerable quantity of sediment, it was found that the amount of moisture in the sediment itself varied perceptibly during the weighing, while the use of the standard eliminated only the variation of moisture in the paper. To correct for this, in heavy samples, the first paper weighed was repeated from time to time and its variation noted, which gave a correction for the moisture in the sediment, reducing all papers to the time when the first one was weighed. In this way the sediment in milligrammes per pint was determined, after which the total sediment was computed as given in "Notes on sediment observations of 1879 at Saint Charles, Mo.," annual report, Missouri River Commission, 1887. Observations for velocity were made in the following manner: Three range lines at right angles to the current and 100 feet apart were located, and their intersections with a base line on shore, perpendicular to them, were permanently marked. Each of these three stations was occupied by an observer with an instrument, signals were given by means of an electrical apparatus, and time was noted by stopwatches, beating one-sixth of a second. Each party was provided with two of these watches. The floats were put out from a floating skiff, and, having started, they were located on the three lines, each observer touching his key as the float crossed his range, at which signal an angle to the float was taken; for the upper range by the lower observer, for the middle range by upper and lower observer, and for the lower range by the upper observer. The skiff, meantime, followed directly in the path of the float and sounded, as nearly as could be judged, at the point where it crossed each range. Both of the watches were started at the first signal, one was stopped at the second, the other at the third signal. The floats were run as nearly as possible at mid-depth, a latitude of four-tenths to six-tenths of the depth being allowed. The discharges were originally calculated by the method of partial areas, using the actual velocity along the float-path. Afterwards, however, the Saint Charles discharges on the Missouri River were recomputed by the graphic method. using normal velocities, as given in "Notes to Missouri River discharge observations," annual report, Missouri River Commission, 1887. The difference in value resulting from these different methods of compensation was, in the main, quite small. APPENDIX F. REPORT ON REDETERMINATION OF STANDARD STEEL TAPE FROM MEASUREMENT OF ENTIRE OLNEY BASE, ILLINOIS, 1887. The work was done by Assistant Engineers O. B. Wheeler, C. V. Mersereau, and R. F. Grady. The preparations, method, tape, and adjuster were the same as described in Appendix A 3 of annual report of this Commission for 1886. The first measurement was attempted on a warm, calm, overcast night, after a day which had threatened rain. The zincs had not been placed, and it required so much time to place them that only the east half was measured. The second measurement was made at night after a day that had been for the most part cloudy and comparatively cool. There was an exceedingly heavy dew, so that tape and thermometer were covered with water. The third measurement was made at night, after an overcast, cool day, with rain at noon, but clearing late in the afternoon. There is open prairie on more than one-half of the 4.10 miles base, and herds of cattle, pasturing there, by rubbing on the stakes, disturbed them, sometimes breaking them down, and it was necessary to go over the line by day before each measurement and re align the nail-heads, and the alignment was never as good again as when first made. The results are, however, very accordant. Two nearly identical values for one tape length (1) from o graduations to the 299-foot graduation are found from the following equations from the measurement of the entire base on the nights of the second and third measurements: (1) 72+ 1+1 foot - 13.0634+27.8342-0.1877 +0.626921623'.6552 - 12′.8545 -2.8342-0.1877+0′.4219216237.6552 from which, by transposing the known quantities, we have: (1) 721-21633'.1119 or l=299′.0752% Mean 1=2997.07525 In these equations the terms are: First term is number of entire tape lengths (o to .299-foot graduation). Second term is fractional tape length between mark on zinc 72 and the mean of three marks on zinc 72 a, which latter marks were from three successive applications of 100 feet portions of tape under a strain of 16 pounds, and for the last 100 feet applying 1 foot of a box-wood scale to the 299-foot graduation to make out the 100 feet. Third term is a distance measured on a straight edge with a box-wood scale from the mean of the three marks on 72 a, above noted, to A west base. (This distance was checked by measuring from the 90-foot graduation of tape to A west base. Fourth term is total distance set forward on zines. (See tables appended.) Fifth term is inclination of tape correction. (See tables appended.) Sixth term is temperature correction to reduce first and second terms to 62° Fahrenheit. (See tables appended.) The second member is the known distance between A east base and A west base. (See page 303, No. 24, of Professional Papers, Corps of Engineers.) A nearly identical value for I is found from the following equations from the measurements of the east half of the base, in which equations the terms are the same as above described, except that for 72a, 100 feet and A west base must be substituted 36 a, 50 feet and ▲ middle base, respectively, and in the second term for 3 substitute 6. from which, by transposing the known quantities, we have (3) 361 10816',5346 or l = 299'.07469 This value for is not independent of that from equations (1) and (2), but, being practically the same, it is preferred to consider the former value, only, in combining with the former determination. The former determination gavel=299'.079 (see last annual report, Appendix A 3°), which combined with that from equations (1) and (2), above, with equal weight, gives 1=2997.077130.00126-neglecting the probable error due from the primary base, which would increase this less than one in the last place. This probable error is one in 240,000. Although this difference in the two determinations is greater than would be expected from the accordance of single values in each determination, yet a probable error of one in 240,000 in this standard of length in a system of secondary triangulation, where, in the reading of angles, a station is occupied but once and readings are made under nearly all conditions of weather, should be considered satisfactory. Nor need the former computations with the larger values for I be recomputed with this new value. I would offer the following reasons for giving the two determinations equal weight: The measurements of 1885 were entirely by daylight on a very favorable, overcast day. while those of this year were entirely by lamplight. By daylight it was possible to see for a certainty that there was no twist in the tape, that the alignment was good, and that the bubble of the level for the horizontal lever arm was well centered, all of which could not be as well attended to by night, and an error in any one of these adjustments would tend to give a smaller value than the true. Also, there was less uncertainty by day than by night in bringing the zero (0) graduation of the tape (which is by the way too coarse a mark) to the mark on the zinc, and there may have been a large personal equation between Messrs. Grady and Sanders in doing this. On the other hand and in favor of the smaller value may be stated that it is from measurements of the entire base, and there was no uncertainty in regard to the hanging of the weight can on the hook of the horizontal lever arm as in the first determination. (See last annual report, Appendix A 3.) A close examination of what this error might have been shows that in one position of the bail of the can the bearing on the horizontal lever arm would have been thrown out 0.27 inch, which would have stretched the 0.00249, and using this as a correction instead of 0′.00085, the result would have been 299′.07746, or essentially the value I would now adopt. The value I would adopt is: 1=299′.07713-0.00126 at 62° Fahrenheit and 16 pounds tension. In conclusion, allow me to state that I have more confidence now than ever before in the steel-tape measurement. But I would introduce one other refinement and take double the time and care in preparing and measuring the base if I wished to insure an accuracy of about one in one million. I would introduce the use of two microscopes of low magnifying power, each to be supported over the zinc strip on a side-arm from its platform, which rests on a horizontal board support for the zinc strip. While the forward observer is numbering his mark on the zine a man brings the rear microscope forward and then guards the forward microscope until the rear observer arrives. The matter of temperature seems now to be under good control, since the extreme difference as given by two thermometers is, for the means, only 0°.15. Very respectfully, your obedient servant, First Lieut. T. A. BINGHAM, Corps of Engineers, U. S. A., Secretary Missouri River Commission. O. B. WHEELER, Temperatures, Olney base. G thermometer is the compared "tape thermometer" of the Mississippi River Commission. R thermometer belongs to the Missouri River Commission, and has been independently compared with the "tape thermometer" for the above corrections.] NOTE.-For corrections to "tape thermometer," coefficient of expansion, and modulus of elasticity, see last annual report, Appendix A3. |