LIST OF APPENDICES ACCOMPANYING THE FOREGOING REPORT. Appendix 1. Effects of Ames Crevasse, March 16, 1891, with diagrams, and *Appendix B of Annual Report, 1891, with its diagrams not then for warded, Col. C. R. Suter. 2. Report on study of velocity of flood travel, with diagrams and subreport of Assistant Engineer Seddon, Col. C. R. Suter. 3. Report on survey of Nonconnah Rocks, with project for removal, Capt. S. W. Roessler. 4. Report of Capt. C. F. Palfrey, secretary of the Commission, with subreports appended, as follows: A.-Secondary triangulation, Keokuk, Iowa, to Port Louisa, Iowa, with tabulated results, description of stations, and plat, Assistant Engineer Stewart. B.-Precise levels, St. Paul, Minn., to Savanna, Ill., field work, reduction, tabulated results, and descriptions of bench marks, Assistant Engineer Ferguson. C.-Precise levels, Duluth, Minn., to St. Paul, Minn, field work and reduction, with tabulated results and descriptions of bench marks, Assistant Engineer Paige. D.-Topographical and hydrographical fieldwork, Assistant Engineer Maltby. E.-Caving banks and state of permanent marks, Assistant Engineer Morrow. F.-Caving banks, areas, and volumes, with graphic summary, As- H.-Discharge measurements, 1891. 5. Report of Capt. S. W. Roessler on operations in first and second dis- A.-Plum Point Reach, Assistant Engineer Nolty. 6. Report of Capt. C. McD. Townsend on operations in third district, with A.-Ashbrook Neck, Assistant Engineer Hider. C.-Louisiana Bend, Assistant Engineer Tollinger. D.-Vicksburg, Assistant Engineer Coppée. E.-Surveys, etc., Assistant Engineer Hider. 7. Report of First Lieut. John Millis on operations in fourth district, LIST OF PLATES ACCOMPANYING THIS REPORT. Diagram showing effects of Ames Crevasse. (Appendix 1, first paper.) Hydrographs, profiles, and gauge relations, Plates I-VIII.* (Appendix 1, second paper. *) Gauge relations, velocities, and time intervals, Plates I-VII. (Appendix 2.) Graphic summary of caving banks. (Appendix 4 F.) Map of Plum Point Reach. (Appendix 5.) Map of Mississippi River near Memphis. (Appendix 5.) Map of part of second district showing levee works by United States in 1891-'92. (Appendix 5.) Map of Ashbrook Neck. (Appendix 6.) Map of Greenville. (Appendix 6.) Map of Louisiana Bend. (Appendix 6.) Map of Lake Providence Reach. (Appendix 6.) General map, fourth district. (Appendix 7.) Works at junction of Mississippi, Red, and Atchafalaya. (Appendix 7.) New Orleans Harbor. (Appendix 7.) Carrollton Bend. (Appendix 7.) Third District Reach, New Orleans. (Appendix 7.) * Not printed. APPENDIX 1. REPORT OF LIEUTENANT COLONEL CHARLES R. SUTER, CORPS OF ENGINEERS, UPON EFFECTS OF AMES CREVASSE, MARCH 16, 1891. ST. LOUIS, MO., June 14, 1892. GENERAL: I have the honor to submit herewith a diagram which exhibits graphically the effects of the Ames crevasse, which occurred opposite New Orleans on March 16, 1891. The diagrams are prepared in a manner entirely similar to those which were submitted to the Commission last year to illustrate the effects of the "Nita" and other crevasses which occurred below Red River in 1890.* The present instance is very interesting, as it was the only crevasse that occurred, and hence there are no complications of the effect produced. Bayou Sara is taken as the upper station on this diagram, as the effect of the crevasse did not extend above Baton Rouge. In the accompanying tabulation are given for each station the maximum effects of the crevasse, that is to say, the difference between the heights which theoretically should have been produced by the maximum reading at Bayou Sara, and the readings which actually did correspond to it. The maximum reading at Bayou Sara occurred on May 1, the crevasse having taken place March 16. An The effect of this large crevasse, which discharged very nearly 100,000 cubic feet per second, died out completely at Baton Rouge, 126 miles upstream, and after running for six weeks it had only relieved the river at Carrollton, 2 miles above, to the extent of 1.9 feet. If the crevasse had not occurred the Carrollton gauge would have reached an elevation of 16.3 feet, or 0.3 foot higher than the actual maximum reading. On the diagram is shown by a full line the gauge relation of 1890; that of 1891, shown by a broken line, is about two-tenths lower at Carrollton. interesting feature of the diagram are the abnormally high readings at all the stations except Carrollton prior to February 16. As this feature shows most prominently at those gauges nearest the "Nita" and other large crevasses of 1890, it seems probable that it was due to deposits caused by those crevasses, which were swept out by the rising river about the date mentioned, lowering the gauge relation to the extent of about half a foot. Table showing lowering effect of Ames crevasse at different points above and corresponding to the crest of the flood at Bayou Sara, May 1, 1891. REPORT OF LIEUTENANT-COLONEL CHARLES R. SUTER, CORPS OF ENGINEERS, UPON THE VELOCITY OF FLOOD TRAVEL ON LOWER MISSISSIPPI RIVER. ST. LOUIS, Mo., June 14, 1892. GENERAL: I beg leave to submit herewith, for the information of the Commission, the results of further study of the velocity of flood travel on the Lower Mississippi. This subject was alluded to in the paper submitted by me under date of August 3, 1888 (published as Appendix A to the Annual Report of Commission for 1891), but for lack of time could not then be elaborated as fully as desirable. The conclusions there stated were derived from a comparison of hydrographs, and of discharges simultaneously measured at different points along the river. The determinations of * Printed in Annual Report, Chief of Engineers, 1891, page 3444. velocity of travel were only approximate, but such corrections as are now made are not sufficient to materially affect the other conclusions embodied in the previous paper. In that paper the gauge relations were constructed by using the approximate time intervals determined as stated above; in the present case the intervals are determined from a careful study of the gange relations themselves. The method used, exceedingly ingenions and interesting, was devised by my assistant, Mr. Seddon, and is fully described in his report to me, a copy of which is appended, and to which reference is made for all details. Briefly stated the method used depends upon the fact that where no tributary increment is received, the readings at a downstream gauge which correspond to those of a gauge upstream should be alike for both rising and falling stages. If they are not, the interval allowed is either too long or too short. By taking trial intervals varying by one-tenth of a day, the smallest fraction that can be reasonably well used in interpolating between the recorded gauge readings, and by studying each rise and fall separately, the interval can be determined to within about a half a tenth a day, or about an hour and a quarter, and this is probably as close an approximation as can well be expected with gauge readings such as are now taken. The sections of river between the main tributaries are taken separately, the uppermost gauge being compared with each of the lower ones in succession. Thus for the section Cairo-Helena, the Cairo gauge is compared successively with those at Belmont, New Madrid, Cottonwood Point, Fulton, Memphis, Mhoons, and Helena. In this way gradually lengthening intervals are obtained, a constant check on the accuracy of the method is furnished, and any tributary effect can be readily determined. As the limit of error in the determination of the interval is a constant one, its percentage of the total will of course vary, and the long intervals are therefore more accurately determined than the shorter ones. In passing from one basin to another it is necessary to choose such periods as are not complicated by tributary increment. This materially reduces the range of the observations, and the interval can not be considered so well determined as in other cases; still, it is not thought that any material error is incurred. The rates of travel as thus determined are given in Mr. Seddon's paper in tabular form and are also exhibited graphically. The rate from Cairo to Red River Landing is 82 miles per day, instead of 75, as previously assumed, and at intermediate points the rate varies both ways from this mean. Below Red River the rate increases enormously and progressively as we descend the river. But for each reach between gauges the rate is constant from the lowest stages to the highest at which the river may fairly be considered to have adjusted its channel; that is, for all stages short of overflow. This constancy of rate, irrespective of stage, and of the variations in the measured velocity, is very puzzling. Attempts to ascertain the mean velocity at different stages over long stretches of river have been made, and the details will be found in Mr. Seddon's paper. The results, however, still show considerable variation with stage, although these variations are less than in the case of individual sections. The conclusion, therefore, seems unavoidable that some form of transmission similar to wave movement is involved. In the deep river below Red River Landing nothing else will explain the very high rate which is noted, and above that point it seems possible that the deep pools may exert a controlling influence in the same direction, especially at low stages when their influence would be most preponderant. At the higher stages, when their influence would naturally be less, the rate does not vary greatly from the mean velocity between Cairo and Red River Landing. In further support of this view, an examination of the tabulation of local rates (Tabulation I) will show that the highest velocities occur on those portions of the river where deep pools preponderate, and the lowest velocities where the mean depths are a minimum. Premising that in such an examination the absolute depth on the bars between the pools need not be considered, but only the relative length and depth of pools and bars throughout the reach considered, attention is especially called to the reaches Cairo-Belmont, Cottonwood Point-Fulton, Arkansas City-Greenville, and St. Joseph-Natchez, as examples of high velocities combined with a marked preponderance of deep pools, and to the reaches of Memphis-Mhoons and Mhoons-Helena as examples of the converse. An apparent exception to this rule is found in the reach Vicksburg-St. Joseph, which can only be explained by the relative shortness and imperfection of the St. Joseph record, which renders the determination of the local rate at this point somewhat uncertain. All this evidence seems to indicate that an increase in mean depth will increase the rate of flood travel, and as a rapid and unimpeded discharge of flood waters is the surest guarantee against excessive flood heights, the bearing of these investigations is thought to be of sufficient importance to justify their publication. Very respectfully, your obedient servant, Gen. C. B. COMSTOCK, President Mississippi River Commission. CHAS. R. SUTER, REPORT OF MR. JAMES A. SEDDON, ASSISTANT ENGINEER, ON THE MOVEMENT OF FLOODS IN THE LOWER MISSISSIPPI RIVER. COLONEL: I have the honor to make the following report on the movement of floods in the lower Mississippi. Up to the bank full stage, or where not complicated by overflow and return flow, and at times when the increments from tributaries are inmaterial, it is apparent that a flood wave as given by lower gauge readings is dependent alone on its upper gauge readings and the law of its movement down the river. It is this law that is sought, and primarily we would see that it might be very complicated, involving changes of shape, dependent on the rates of rise and fall, on different rates of movement at the different stages, and on the existence of reservoir capacity in the river; also the rate of movement might vary, from time to time, and might have different values for floods of different heights. The great effect of these possible complications is found by platting the gauge relations with constant trial time intervals for successively lengthening reaches. If the relation gives the rise and fall coinciding, or following the same line, when platted with a determined constant interval, it shows that the above complications are then not perceptible in that reach; and by extending to longer and longer reaches we get a measure of the degree of accuracy with which we may say that flood movement is free from them. At the same time, by extending the reach we identify the beginning and growth of the effects of tributary increment, which frees the irregu larity below from suggesting a reasonable suspicion of change in the law of flood movement. These conclusions rest on the following reasoning: Up to the crest on both the rising and falling stages each point in the flood wave at the upper gauge must have its period, after which the same value of discharge, and hence the true equivalent of its gange height would be found at the lower gauge; now, if in platting a trial gauge relation our assumed time interval were less than the true periods, or too short, then on the rising stage the lower gauge values taken, would not have had time to rise to their true equivalent value, and would be too small in proportion to the rapidity of the rise. In the same way on the falling stage the lower gauge values would not have fallen enough and would be too large in proportion to the rapidity of the fall. The trial relation would therefore plat as in Fig. 1, where the rising and falling stages are indicated by the directions of the arrows. By the same reasoning we see that if the trial interval had been taken too long the lower gauge values would have been too large on the rising and too small on the falling stages, giving a trial relation like Fig. 2; and if between the trial intervals of Figs. 1 and 2 we find an interval that causes the relation to plat as in Fig. 3, the rising and falling stages coinciding throughont, we must conclude that the flood wave passed down as a whole in this constant time interval, with no perceptible change of shape. It may be noted here that the position, inclination, and straightness of this line in Fig. 3 in no way affect the above conclusion, nor will these questions be considered in this report, as they form the basis for the independent studies of the relations of discharge curves along the river, and changes in channel efficiency. Now the degree of accuracy with which this time interval may be determined depends on the rate of rise and fall, and on the accidental errors of the gauge readings; and while the short distance gauge relation in itself might show that there was no phasing of the wave, or change of its interval, perceptible in the short reach, yet it would not follow that these effects, growing with the distance might not be material in the longer reach. But when we extend the reach by keeping the same upper gauge and taking a lower gauge farther down the river, and finding the new interval as before, see the coincidence of the rising and falling stages recurring with an equal precision; we have at once a measure of our accidental errors, and the answering argument that divergence does not increase with distance. By further extending the reach and still finding the flood reproduced in ratio on the lower gauge with the same precision, we are, with a high degree of confidence led to accept as a fact, that the law of flood movement is practically an unphasing movement of the wave as a whole at constant rates from point to point down the river. The permanence of the interval and its constancy for floods of different heights, is of course determined by extending the investigation to other floods. Plates I to V present this study of the gauge relations. Plate I gives the relations in the reach from Cairo to Helena. In this Cairo is uniformly taken as the upper gauge, and the reaches for which the gauge relations are platted are, successively, Cairo to Belmont, Cairo to New Madrid, and so on to finally the reach Cairo to Helena; as in all former gauge relations the dates uniformly refer to the actual time of the upper gauge readings. Group 1 shows a small flood platted with trial intervals varying by one-tenth of a day, and from this the true interval is easily selected to the nearest tenth. This only leaves a limiting error in the interval of, say, 0.05 of a day, or something over one hour, and it is useless to attempt to reach a greater accuracy than this with the data, on account of accidental errors and probable irregularity in the actual time of reading the gauges. The bending away on November 13, 14, and 15 in the direction of too large gauge readings from Fulton down is an effect of tributary increment and coincidence with the commencement of a rise in the St. Francis River. Group 2 gives the relations for a higher flood. The Cairo flood is shown exactly reproduced down to Memphis; Helena shows a small tributary increment on the falling stage (at the maximum about 20,000 cubic feet per second). The case of Mhoons deserves a special notice; on January 1 (January 4 at Mhoons) the gauge reader commenced taking the readings on a temporary gauge and continued so to May 23, when he was found to be about 2.2 feet too low. From the gauge relation it is evident that a little over 2 feet of this error was made in the first three days, and the readings might be very safely so corrected. But if this error had been distributed over the whole time, as is often done, it would have given a perfectly incomprehensible condition of the river at Mhoons for about five months. This instance serves not only to point out the necessity of correcting gange errors by the gauge relations, if any corrections are to be made, but also the care that must be taken in the study of data already collected in order not to be led into false conclusions from similar cases. Group 3 gives the relations for a very low rise, and group 4 for the very sharp rise in the fall of 1890. In this last the gauge reading 19.9 Cairo gauge, on the falling stage, is clearly 1 foot in error in its Cairo reading, and has been so corrected and shown as a dotted line. Plate II gives other groups of gauge relations from Cairo to Helena. Plate III gives gauge relations from Helena to Vicksburg. In this reach we have the tributary increments from the Arkansas, White, and Yazoo rivers, which prevent a satisfactory study of the intervals throngh it as a whole. Group 1 shows a study of the mouth of White River-Arkansas City interval, with various trial times; group 2, a study of the intervals with trial times from Arkansas City to Vicksburg; group 3 another flood from Arkansas City to Vicksburg; and group 4 a study of the intervals from Helena to mouth of White River. In this reach, from Helena to mouth of White River, it is very difficult to find periods in which the movement of floods can be closely studied on account of the large tributary effects. Plates IV and v give the study for the reaches Vicksburg to Red River Landing and Red River Landing to Carrollton, respectively, and with what has preceded they need no general explanation. In the Vicksburg-Red River Landing reach the special tendency to change of plane at Red River Landing, considered in former reports, complicates the study of the interval. An instance of this change of plane is shown in group 5, at Red River Landing, between the rising stage of early in November and that of the latter part of December. In the Red River Landing-Carrollton reach the smallness of the range at College Point and Carrollton make the determination of the interval less accurate. Also there are upstream effects where the interval is reversed, as in the latter part of the falling stage, September-October, 1890 flood, as well as small apparently erratic changes of plane, as that separating the falling and rising stages in the flood of November-December, 1890, all of which add to the difficulties of determining the interval. But notwithstanding these difficulties, the marked acceleration to the movement of the flood as it passes down the reach is well shown; and there is little doubt but that the old interval used, of one day from Red River Landing to Carrollton, if |