The purpose of this paper is to present a brief sketch of the development of this hindered-settling “sorting” classifier, but primarily to show the actual results obtained in practice with the classifier working on the Butte copper ores at the Boston & Montana Concentrator at Great Falls and the Washoe Concentrator at Anaconda. The writer has endeavored to be as brief as possible, giving only the essential figures showing the efficiency of the classifier and leaving considerable interesting information to be derived by the reader from the statistics given.

The paper deals with the practical work of the classifier under the following heads:

1. Desliming 4.0-mm. primary feed, overflow from 0.07 to 0.0 mm.
2. Desliming 1.25-mm. secondary feed, overflow from 0.07 to 0.0 mm.
3. Desliming Huntington Mill discharge, overflow from 0.07 to 0.0 mm.
4. Classifying 2.5-mm. deslimed secondary feed, overflow from 1.00 to 0.07 mm.
5. Classifying 2.5-mm. deslimed primary feed, overflow from 0.75 to 0.07 mm.
6. Classifying 4.0-mm. deslimed primary feed, overflow from 0.75 to 0.07 mm.
7. Classifying middling-tailing product from roughing tables, overflow coarse tailing.
8. Classifying 2.5-mm. trommel undersize, not deslimed, overflow from 0.75 to 0.0 mm.
9. Remarks on the inner feed cone.

In the following pages all overflows are given in quartz dimensions; i. e., by a 0.75-mm. overflow is meant an overflow whose maximum particle is a 0.75-mm. quartz grain. By the term “slime” is meant all material which will pass through a 200-mesh screen whose average opening is 0.07 mm.

The development of the classifier was commenced in the fall of 1910, and was primarily the result of a suggestion by Dr. Robert H. Richards, that a conical tank equipped with a pulsating current would deslime the feed to his direct pulsator classifier then installed in No. 4 Section of the Great Falls Concentrator. The success of this arrangement in removing the slimes from ore pulp, together with the demand for a deslimed feed to the Hancock jigs, led to a series of experiments initiated by A. E. Wiggin in the fall of 1910. The final result of these and later experiments has been the production of the Anaconda hindered-settling classifier, which has as its chief assets the three leading factors to be reckoned within any machine: namely, simplicity of construction, large tonnage, and high efficiency.

In criticizing the work of any classifier it must be remembered that we are dealing with grains of material of every conceivable size and shape from flat to round grains, and with varying specific gravities, from the purest and lightest gangue to the purest and heaviest mineral grains, together with the innumerable combinations of the same. If a classifier were mechanically perfect it still would be impossible to separate a large flat grain of gangue from a much smaller rounded grain of rich middlings.

Fig. 1 shows the first form of the classifier constructed. The early experiments proved that the pulsating current was of no advantage in this type of machine, hence it was abandoned for a plain rising current. The classifier consisted of merely a vertical pipe for a sorting column with a superimposed inverted truncated cone, the hydraulic water being admitted at one side near the bottom of the sorting column at right angles to the spigot discharge.

The experiments immediately following consisted chiefly in varying the dimensions of the sorting column and the conical top, which will be referred to hereafter as simply the “top.”

The object sought in these earliest forms was merely a deslimer which would give an overflow practically all through 200 mesh and at the same time give a clear spigot free from slime. Many forms were tried. Fig. 2 shows one of the intermediate stages.

The hydraulic water was now admitted at 180° to the spigot discharge by means of a 90° elbow at the center of the sorting column. The work of this particular stage, while an improvement over earlier forms, was very crude when compared to the present form of deslimer.

Tons treated per 24 hr……………………………………………………108
Per cent, total feed overflowed……………………………………….12.0
Percent, solids in overflow on 200 mesh…………………………45.7

Laboratory Test on Model Size Three-Pocket Unit

Fig. 3 is a sketch to scale of a unit of three pockets in series, built for the purpose of determining in the laboratory the relative velocities of rising currents required in the sorting columns to make the proper separations other than at 0.07 mm., and particularly at 1 mm., as this point had been decided upon as the point of division between Wilfley table feed and Hancock jig feed in the mill flow sheet. The velocities shown on the sketch are the theoretical velocities and not the actual values. The actual velocities could not be determined in the experiment because of the lack of suitable means for measuring accurately such small quantities of water as were used by the laboratory classifiers. The feed to the unit consisted of the undersize of a 5-mm. round-hole trommel screen.

The sand was fed dry to the first classifier by means of an automatic shaking feeder. The rising current in the sorting column of the first classifier was adjusted so that the dividing line between the first and second pockets was at about 1 mm. It was attempted to keep the overflow of the final pocket entirely through 200 mesh. The three spigot products and the overflow of the final pocket were screen sized, and each size was sorted into free mineral grains and quartz or middlings grains. The coarser sizes, from 4 mm. to 2 mm., were sorted by hand, while the finer sizes, between 2 mm. and 1 mm., were treated in a sorting tube, and the product below 1 mm. was concentrated on a laboratory size Wilfley table.

The amount of free mineral in the overflow (through 200 mesh) was estimated. The results of the test are shown in Table I and Fig. 4.

The classification chart, Fig. 4, which brings out the degree of classification quite distinctly, was worked out by C. R. Kuzell.

The free mineral overlaps on three screen sizes, 50, 60, and 90 mesh. The quartz grains in Spigot 1 which are finer than 1 mm. were fairly rich middlings grains whose specific gravities were greater than for those grains in Spigot 2 on 60 mesh, the latter being for the most part grains of quartz containing little or no mineral, and probably the bulk of these were flat grains. The feed was distributed as follows:

The Adoption of the Constriction Plate

Up to this point no form of constriction had been used in the sorting column; and although a teetering mass had been observed in the column at times, there is no doubt in the writer’s mind but that this was due to overcrowded free-settling condition and not to true hindered settling. To produce strictly hindered-settling condition a constriction was adopted in the form of a thin plate of steel with a central circular opening. The result was to produce a teetering mass above the opening in the plate (sorting column), which will be called the teeter chamber. Later the depth of the sorting column was increased by using a nipple 2 in. long. The object of using a deeper sorting column was to do away with local eddy currents and to take up more or less the slight fluctuations in the hydraulic pressure. This lengthened sorting column increased the efficiency considerably, for at this point is determined whether a grain shall rise into the overflow or pass out the spigot. The teeter chamber had now been increased to 12 in. in diameter, and this diameter was adopted as a standard size. From the theory of hindered settling it is seen that the ratio between

the cross-sectional areas of the teeter chamber and the sorting column is of the utmost importance. If the ratio is too small, free-settling classification is the result; if too large, the classifier will bank or choke up; while intermediate between these two ratios is one which will establish equilibrium between the teetering mass and the discharge through the sorting column. With this ratio established for a given condition, the classifier will maintain a bed of quicksand above the constriction, with

a constant discharge through the spigot. Numerous interesting experiments have been followed out along this line, with the result that the ratios given in the following pages have been found best. The single opening has been discarded for numerous smaller openings, with the idea of equalizing the rising current throughout any given cross-section of the teeter chamber.

The present form of the classifier is installed in six of the eight sections of the Great Falls Concentrator and the remodeled Section No. 1 of the Anaconda Concentrator, of which it may be said to form the nucleus.

The Classifier in Actual Practice

The classifiers may be divided into two groups, according to the work performed: namely, desliming classifiers and table-feed classifiers.

Desliming Classifiers

Fig. 5 gives a cross-sectional view of a standard 7-ft. deslimer, showing the essential point of construction.

Fig. 6 is a photographic view of a deslimer in use at the B. & M. Concentrator, Great Falls. These deslimers are used for desliming pulp from 4 mm. down and have proved very efficient. One particular feature which has been found to hold in practice is that when desliming fine pulp (2.5 mm. or under) no rising hydraulic water is used in the classifier; i. e., the only fresh water used in the deslimer is that used to supply the spigot, which is usually about 50,000 to 90,000 gal. per 24 hr. However, in desliming coarser feed some rising water is generally used. The B. & M. practice is to use a constriction of one 6 in. diameter by 2 in. long pipe, giving a ratio of 3.9. The Anaconda practice is to use a number of smaller pipes with a ratio of 4.6. Which is the better cannot be said for conditions of practice are considerably different at the two mills. The shortest and best way to show what the classifiers will do in practice is to quote actual figures, obtained under different conditions.

If fed under ideal conditions a 7-ft. deslimer will handle 250,000 gal. of pulp with 200 tons of solids, giving an overflow of not over 2 per cent, of total solids in the overflow on 200 mesh. The spigot will be perfectly clear and very dense.

Desliming 4.0 mm. Primary Feed, Overflow from 0.07 to 0.0 mm.

“Primary” feed contains all the original mine fines and no finishings roll product. Of necessity it is a very rich feed and contains a large amount of free mineral.

Following are figures representative of the work done at Great Falls on the above class of feed. The sum of the plug and the overflow is taken as the feed, which in all cases checked the actual feed sample.

Gallons of water consumed per ton of feed, 905.
Diameter of spigot discharge, 1.25 in. Spigot clear.
Rising velocity through constriction = 226 mm. per second.
Ratio of cross-sectional area of teeter chamber to sorting column = 3.9.
Constriction opening = one pipe 6 in. in diameter, 2 in. long.

The overflow is shown as containing 7.9 per cent, of total overflow on 200 mesh, which is no doubt due to excessive use of hydraulic water. Under test condition this same deslimer has shown 2.3 per cent, on 200 mesh, which is normal. However, as these and all figures following, except where otherwise stated, are actual practice figures and not test figures, the writer will explain from his personal observations and experience with the classifier such abnormal figures as might be misleading to the reader unfamiliar with the classifier.

The same deslimer when treating a little coarser feed gave as follows:

Gallons water consumed per ton of feed, 840.
Diameter of spigot, 1.25 in. Spigot,

The overflow screen sized 6.6 per cent, of total solids on 200 mesh. The water consumption is low, but could have been made considerably lower by running a denser spigot.

Desliming 1.26 mm. Secondary Feed, Overflow from 0,07 to 0.0 mm.

In this case the “secondary” feed consisted of the undersize of 1.25 by 12 mm. punched-plate trommels screening finishing-roll product. The feed was very dense and afforded excellent feed for a deslimer.

Gallons water consumed per ton of feed, 409.
Rising velocity through constriction, 84 mm. per second.
Diameter spigot 0.75 in. Spigot contains trace.

The above results were obtained from one of the secondary deslimers in use at Anaconda running under normal conditions. The constriction opening in this case was 11 1.5-in. nipples 3 in. long giving a ratio between the area of the teeter chamber and the sorting column of 4.6.

Desliming Huntington Mill Discharge, Overflow from 0.07 to 0.0 mm.

The feed was the discharge of one or more 5-ft. Huntington mills clothed with 1 by 12 mm. punched-plate screens. The practice was to deslime the above feed and treat the spigot on finishing Wilfleys. The writer has observed tables fed in this manner handle close to 50 tons per 24 hr., and make 80 per cent, of the feed into 0.45 per cent. Cu tailings. One of the reasons for this splendid work is the distinct valley produced on the table as the hindered-settling effect of the classifier, resulting in the clean separation of free mineral and gangue grains. Oftentimes this valley was from 10 to 12 in. wide and extended the entire length of the diagonal division line between the mineral and the gangue. Very little middlings was made on these tables. I shall give two sets of figures under this type, showing a medium volume of feed and an excessive volume of feed.

Gallons water consumed per ton of feed, 590.
Diameter of spigot, 1 1/8 in. Spigot, clear.
Constriction opening, one 6-in. pipe 2 in. long.
Constriction ratio, 3.9.
Rising velocity through constriction = 64 mm. per second.

The excessive fine material in the feed, 20.5 per cent, through 200 mesh, is due to the use of roll-product slime as sluicing water in the mill aprons.

Following are figures showing the work of the same deslimer under a much smaller volume of feed pulp:

Gallons fresh water consumed per ton of feed, 1,000.
Same size spigot and constriction as above.
Rising velocity through constriction = 41 mm. per second.

In desliming classifiers the volume of the feed pulp is the most important factor. Experiments have shown that with a deslimer with an 84-in. diameter top the pulp volume should never exceed 250,000 gal. to make an overflow 100 per cent, through 200 mesh. The reason for this is apparent.

Effective cross-section near overflow level = 5,290 sq. in.
Overflow = feed volume plus rising water.
Rising water = none.
Overflow = 250,000 gal. per 24 hr.
250,000 x 231 x 25.4/86,400 x 5,290 = 3.2 mm. per second rising current near overflow level.

Richards gives the free-settling velocity of quartz whose average diameter is 0.0747 mm. as 3.57 mm. per second. The feed must be constant, however, if we expect to use no rising water in the classifier.

From the foregoing, the following conditions are most favorable for desliming 2.5 mm. pulp:

1. Maximum feed volume, 250,000 gal.
2. Maximum feed density, 17 per cent, solids.
3. Maximum tonnage, 250; average tonnage, 200.
4. Rising velocity at overflow level, 3.2 mm.
5. Rising velocity through constriction varies with the size of material, from close to zero to 40 mm.

For material coarser than 2.5 mm. the density of pulp through the sorting column (constriction) must be lower, to prevent the possibility of choke-ups. As a consequence a rising current must be maintained in the sorting column or slime will enter the spigot. The finer pulp (2.5 mm. or below) seems to act as a filter for the slime feed and oftentimes results have been reported showing the extraction of deslimed water from the

feed and the addition of the same to the spigot. This can only be accounted for by the filtering effect of the teeter chamber. With feed above 2.5 mm. this phenomenon has seldom been observed.

Table Feed Classifiers

The second division of these classifiers are called table-feed classifiers because their function is to overflow a product which may be best concentrated on some type of shaking table. The chief use to which this classifier is put is to overflow a 0.75 mm. table feed, either primary, or secondary, or mixed.

Fig. 7 is a cross-section of one of the table-feed classifiers, of the type installed at the Washoe Concentrator. Unfortunately, no figures could be obtained showing the work of these classifiers in practice. At the B. & M. Concentrator an inner cone has been added to the original design and has found universal use in this type of classifier. The development of the feed cone was begun in May, 1912, the first classifier with an inner

cone being installed, I believe, on May 9, 1912. Although we have no figures showing the classification of the Anaconda classifiers at the Washoe Mill, Fig. 8 gives a very clear conception of the actual work done by them. Dr. Richards states in his paper, The Development of Hindered- Settling Apparatus, that the advantage of hindered-settled feed on a shaking table is the production of a valley between the mineral and the sand, thereby aiding the separation of the different products.

In Fig. 8 we see this valley as obtained in actual practice. The photograph was taken after shutting off the feed and dressing water simultaneously and then shutting down the table approximately 10 sec. after the feed and water were cut off. The table was handling from 15 to 18 tons of feed per 24 hr. The tailings at the deepest point were 0.75 in. deep, the concentrates 0.25 in. deep, and the middlings about 1/16 to 1/8 in. deep. The tailings will probably assay 0.30 per cent, of copper. The feed to the table ranged from 0.75 to 0.07 mm. and was overflowed from the classifier shown in Fig. 8. The apparent broken spot on the Wilfley is due to reflections.

Classifying 2.5 mm. Deslimed Secondary Feed, Overflow from 1.00 to 0.07 mm

The feed was deslimed rolled product through 2.5-mm. round-hole trommel. The overflow was heavy and classification not very close. The classifier was one of the type shown in Fig. 9, no inner cone being used.

Gallons water consumed per ton. of feed, 1,165.

Fig. 9 shows the overflow of the above classifier.

Classifying 2.5 mm. Feed, Overflow 0.75 mm

The classifier was equipped with an inner feed cone (Fig. 10). The feed consisted of 2.5-mm. round-hole trommel undersize, deslimed.

Gallons water consumed per ton of feed, 1,100.
Diameter of spigot, 1.25 in.
Constriction opening, 12 2-in. pipes 2 in. long, or 40.0 sq. in.
Ratio of area of teeter chamber to sorting column, 2.81.
Rising velocity through constriction = 316 mm. per second.
Rising velocity at bottom of annular space = 144 mm. per second.
Rising velocity at top of annular space = 122 mm. per second.

The figures show this classification to be excellent, although the classifier was designed to receive a feed of 250,000 gal. per 24 hr., whereas it is receiving only 175,000 gal. In the spigot fully 80 per cent, of the fine material, through 0.707 mm., is free mineral grains, which we would expect in any spigot, even the perfectly classified.

A second set of figures shows the work performed on a feed containing more fines.

Gallons water per ton of feed, 1,050.
Spigot, 1.25 in. in diameter.
Constriction area, 40 sq. in.
Constriction ratio, 2.81.
Rising velocity through constriction = 373 mm. per second.
Rising velocity at bottom of annular space = 152 mm. per second.
Rising velocity at top of annular space = 129 mm. per second.

The work done by this classifier, as shown by the two sets of figures, may be said to be the best obtained with any type of the Anaconda classifier. The separation at 0.75 mm. is remarkable and shows the possibilities of the machine. This particular unit is used at the B. & M. Concentrator to furnish Hancock jig feed and finishing-table feed. Figs. 10 and 11 show the overflow and spigot of this installation as in use to-day. Note the appearance of the overflow in the photograph. It is very even except for the agitation produced by the inner cone throwing backwards into the overflow, material which is carried up the outside of the cone.

Classifying 4.0 mm. Original Feed, Overflow 0.75 mm

The feed was deslimed original material, being the undersize of 5.0 mm. round-hole trommels. Classifier equipped with feed cone.

Gallons water per ton of feed, 445.
Spigot diameter, 1.25 in.
Constriction area, 40 sq. in.
Constriction ratio, 2.81.
Rising velocity through constriction = 136 mm. per second.
Rising velocity at bottom annular space = 109 mm. per second.
Rising velocity at top annular space = 92 mm. per second.

Classifying Middlings-Tailings Roughing-Table Product, Overflow Coarse Tailings

At the present time there are no classifiers in the B. & M. Concentrator overflowing tailings. This is due chiefly to the lack of head room, as the mill is old and not designed along modem lines.

The B. & M. system of concentration as installed at the Washoe mill includes four of these classifiers in a 1,500-ton unit, although two of them only are used most of the time. Fig. 12 is a view of one of the tailings classifiers in use at the Washoe mill. The classifier in this case is discharged through the lower spigot direct to Huntington mills. The feed is the middlings-tailings product from the roughing tables and the overflow averages 0.45 per cent, of copper. Two spigots are provided for the classifier, the original idea being to use one in case the other choked. However, it has been found unnecessary to provide the second spigot, the later types being cast with but one (see Fig. 7). The following data are not taken from regular practice but are taken from a several days’ test run and suffice to show what the classifier will do under this class of feed. The so-called middlings-tailings consist of true middlings grains and coarse tailings grains with no free mineral. The test was carried out according to the following flow sheet:

The numbers applied to the classifiers are arbitrary and are used merely for convenience. Fig. 15 shows the cross-section of the No. 2 classifier. Fig. 13 shows the No. 7 classifier, and Fig. 14 shows the distribution of products on the roughing Wilfley.

The No. 7 classifier used 127,000 gal. of fresh water, or 350 gal. per ton of feed. As seen from the sketch, Fig. 13, the classifier was a very small

one, handling only 36 tons of feed. Still, there is every possibility of building a larger machine with just as great an efficiency. The most essential feature of this system is that the roughing Wilfley must produce a middlings-tailings product absolutely free of fine mineral. The writer has seen roughing Wilfleys at the B. & M. Concentrator doing very good work on 130 tons of 2.5 to 7.5 mm. feed.

The screen sizing figures are self explanatory. The screen sizing of the rough-concentrates shows the segregation of “through 0.841 on 0.500 mm.” material, amounting to 57.8 per cent, of the total rough concentrates, or 18.2 per cent, of the total feed to the table. It may be well to state here that the No. 2 classifier preceding the roughing table, while doing good work, was one of the earliest fitted with the feed cone. One of the present-day cone classifiers would eliminate a great percentage of this fine material. Some interesting tests have been made on this No. 2 classifier and the results follow. This particular No. 2 was the pioneer of the present inner feed cone classifier as used at Great Falls.

For the method of designing the Anaconda classifier reference is made to a paper before this meeting by E. S. Bardwell, The Application of Hindered Settling to Hydraulic Classifiers.

Classifying 2.5 mm. Roll Product (Not Deslimed) in No. 2 Classifier

The following test figures are actual practice figures. The classifier was not altered or adjusted in preparation for the test. Samples were taken on two days practically at 1-hr. intervals and each sample was screen sized and sorted into free mineral grains, true middlings, and tailings grains. The classifier was fed direct from the trommel undersize, taking all of the feed at this point of a 500-ton section. The primary object of the test was to show the work of the classifier under varying

feed. The feed varied during this run from 90 to 250 tons per 24 hr., although the pulp volume was fairly constant. The tabulations show remarkably good classification under this changing feed.

The classifier was provided with a 1.25-in. spigot and used an average of 124,000 gal. of total fresh water per 24 hr., this being determined as an average of many samples.

The following tabulations and classification charts tell the story at the time of each different sample more clearly than words.

The symbols used in the tabulation are as follows:

V1 = Rising velocity through constriction in millimeters per second.
V2 = Rising velocity in teeter chamber in millimeters per second.
V3 = Rising velocity at bottom of annular ring in millimeters per second.
V4 = Rising velocity near top of annular ring in millimeters per second.

A = Per cent, of variation of feed pulp, gallons per 24 hr. above and below average.
B = Per cent, of variation of feed pulp solids, pounds per 24 hr. above and below average.
C = Average diameter in millimeters of quartz in plug discharge.
D = Average diameter in millimeters of mineral in plug discharge.
E = Ratio of C and D.
F = Tons of plug discharge per square inch of constriction opening per 24 hr.
G = Tons of plug discharge per square inch of cross-section of teeter chamber per 24 hr.

Particular attention is called to the consistency of classification, and especially the low water consumption.

Richards’s velocity for full teeter of 1.37 mm. grains is 53 mm. per second. That obtained in the classifier was 39, but this calculation was made assuming the same density in the teeter chamber as in the sorting column. As a matter of fact, had this density been determined, the rising velocity in the teeter chamber would have checked Richards’s velocity. The low consumption of water is due to a great extent to the inner cone.

Inner Cone

The inner cone, in the writer’s opinion, is a great improvement over the type of classifier without the cone. The chief function of the cone is to produce a constant rising current under all conditions of operation. It also acts as a feed box and reduces liability of choke-ups. By utilizing every drop of feed water for the production of the rising current, the amount of rising fresh water is not much greater than that required to produce the quicksand or full teeter condition in the teeter chamber. This fact not only reduces considerably the total fresh water consumed, but produces a much better average classification than a classifier without the cone. In a classifier with no inner cone, if the solids in the feed pulp decrease the teeter chamber may disappear entirely and we have simply a free-settling classifier with considerable fine material being discharged through the spigot. But if in an inner cone classifier the density decreases, i. e., the volume remains constant but the solids decrease, we still have the same rising current, which has a tendency to lift out all of the fine material and a good part of the coarser feed. The inner cone has given best results when so designed that the velocity at the top of the annular space is the same as the free-settling velocity of the maximum quartz grain we wish to overflow and the velocity at the bottom of the annular space is approximately 25 per cent, greater. Some particles are drawn up into the space which eventually settle down and are discharged through the spigot.

As seen in Fig. 10, the overflow is constantly under a head of from 5 to 6 in. in the inner cone. Any tendency to bank, results in the increase of this head and a cleaning up of the banking material. However, banking or choking is rare with the inner cone type.

The advantages of the inner cone may be summed up as follows:

1. A uniform close classification is obtained.
2. The water consumption is decreased.
3. The liability of choke-ups is decreased.
4. Boiling in the overflow is prevented, thereby preventing coarse material being thrown over.

The chief disadvantage is the requirement of 8 in. of extra head room.

In any classifier or classifier system, the most important demands to be satisfied are: (1), flexibility of each unit and of the classifier system as a whole; (2), maximum tonnage; and (3), minimum water consumption. Let us see how the Anaconda classifier answers these fundamental requirements.

(1). The Anaconda classifier system is remarkably adapted to meet any given set of mill conditions. The classifier units can be operated under the direct system or under the indirect system. In the direct system the slime is the product of the last classification. In the indirect system the slime is the first product removed. The direct system is best adapted to conditions where mill head room is exceedingly valuable. The total pulp, usually a screen undersize, is fed directly into a table-feed classifier which overflows all material from a given size, usually 0.75 mm., down to 0.0 mm.

The overflow is easily laundered or pumped to some point where it can be further classified into slime and one or more classified table feeds. The valuable point here is that the second and third classifiers in this series can be placed immediately preceding the particular concentrating machine best adapted for the treatment of each product, allowing the practice of running very dense spigots as feed to the concentrators, which are usually some type of shaking table. Reference to Fig. 8 will show the manner of separation of such a feed into concentrates, middlings, and tailings by a Wilfley table. The concentrates are very clean, usually containing from 9 to 15 per cent, of gangue; the middlings are very small in tonnage, while the tailings are very dense and assay usually from 0.30 to 0.40 per cent, of Cu.

The indirect system first removes the slime (98 per cent, through 200 mesh), which can be segregated by itself in a plant equipped for slime treatment. The density of the slime produced in the indirect system is from 160 to 200 per cent, as dense as in the direct system. This is due to the fact that no hydraulic rising water is required for the separation of the slime alone. The spigot of the deslimer is fed directly into a table feed classifier, which affords to the classifier an absolutely constant volume of feed pulp. This fact is the most important feature of the indirect system. A variable volume of feed pulp to a deslimer is immaterial provided the deslimer is designed to handle the maximum volume. A variable feed volume to a table-feed classifier is serious in that the rising velocity varies proportionately with the feed volume.

As to the flexibility of each unit, the tabulation following Fig. 10 shows the results of a variable tonnage. This particular unit, which is one of the earliest inner cone types, received as its feed the total undersize of two 2.5 mm. round-hole trommels. This feed was far from constant, varying from 90 to 255 tons per 24 hr., and yet the classifier maintained practically the same degree of classification under all conditions. A variation of 40 per cent, of its rated tonnage, either way, is not serious in an Anaconda classifier. The hydraulic water is the only adjustment necessary, and this is only a slight one.

The effect on the degree of classification due to variations of the percentage of fine material (mine fines) in the feed is minimized in the Anaconda classifier. In any mill which is not equipped with a central crushing plant the variations in the percentage of mine fines in the original ore and even the percentages of fines due to crushing are considerable. To perform consistent work a classifier must be able to absorb these fluctuations in the quality of its feed. On pp. 296 and 297 are shown two complete sets of figures illustrating to a certain extent how well this function is performed by the classifier.

A comparison of these two sets of figures shows the same quality of work on the rich fine feed as upon the coarser and more impoverished feed. Both sets of figures were obtained under actual operating conditions.

(2). The maximum tonnage of the classifier is limited only by practice. In Montana it is the custom to design each unit to treat from 150 to 350 tons per 24 hr., depending upon conditions of practice. Table-feed classifiers seldom treat less than 250 tons, while deslimers working on very fine feed, 2.0 mm. or under, will average from 180 to 250 tons. Deslimers treating material whose limiting size is larger than 2.0 mm. have a capacity of from 250 to 400 tons per 24 hr. The large tonnage handled by a single unit is one reason for its great flexibility.

(3). The consumption of water by these classifiers is remarkably low. In all desliming Anaconda classifiers practically no rising hydraulic water is used, leaving only that which is added in the spigot to be charged against the classifier. However, the addition of water in the spigot reduces the consumption of water in the subsequent classifier. The deslimed overflow of the table-feed classifier is dewatered, and the water thus removed is used as hydraulic water in other classifier units of the system or as dressing water for concentrating machines. Thus the total water consumption for a system of classifiers is less than is shown by the sum of the individual units. The average water consumption in practice is as follows:

Desliming classifiers:—600 gal. per ton of feed.
Table-feed classifiers:—450 to 900 gal. per ton of feed.
The relatively large consumption of water by the deslimer is due to operating with a spigot less dense than is used in table-feed classifiers.

Fulfilling as it does the requirements of flexibility, efficiency of classification, high tonnage, and minimum water consumption, together with simplicity of construction, the Anaconda classifier is a great stride in the progress on modern ore concentration.