[1] The New York Continental Jewell Filtration Company, plaintiff herein, and hereinafter to be called the filtration company, brought suit against the city of Harrisburg, defendant in the court below and appellant here, alleging infringement of reissue letters patent upon a filter and of letters patent upon a method o'f purifying water, both alleged to he the invention of O. H Jewell, and praying for an injunction and accounting. Defendant denied the validity of the patents and infringement of them. After trial, it was-decreed by the District Court that the patents were valid, and that the claims had been infringed. Injunction and accounting were ordered,.. and from such decree the city of Harrisburg has appealed.

Claim No. 2 of the reissue patent involved is as follows:

“(2) A filter consisting of a filter tank, a pure-water pipe communicating with the lower portion thereof, a granular filter bed in said tank, the lower portion of said filter bed being compacted more closely than the upper por*367tion thereof, means for maintaining a partial vacuum in said filter bed during filtration, and means for reversing the flow of water through said filter bed, substantially as described.”

The three claims involved in the method patent are as follows:

“ (1) The method of purifying water, which consists in passing the impure water through1 a granular filter-bed having an exposed filtering-surface, and at the same time applying suction from below, substantially as described.

» “(2) The method of purifying water, which consists in passing the impure water downward through a granular filter-bed of gradually-increasing compactness from the top downward, and at the same time applying suction from below, substantially as described.

“(3) The method oí purifying water, which consists in passing the impure water through a granular filter-bed having an exposed filtering-surface, and at the same time extracting from the wafer by suction and retaining within the filter-bed a greater or less portion of the air contained in the water, substantially as described.”

The invention disclosed by the patents in suit has reference to atmospheric filters, filters in which, by the creation of a partial vacuum within or below the filtering material, atmospheric pressure above the liquid to be filtered is rendered effective to force it through the filtering material.

There are two general types of such filters: The slow sand, or English, filters, and the rapid, mechanical, or American, filters. Of the latter, some use pressure, and some use gravity; the Jewell filter belonging to the gravity class. Essentially the bed of each type is of a granular material, usually sand, exposed to the air and resting upon gravel; the gravel containing a' system of small pipes which collect the filtered water and deliver it into a large outflow pipe connecting with the distributing system. In both types, less water actually passes through the bed than its full capacity; the rate of flow is determined, not only by the type, but also by such conditions inter alia as the turbidity of the unfiltered water, the amount and kind of bacteria, and the quantity of water needed for use. The quantity is regulated by rate-controllers, devices which are in practically universal use. The slow sand filters deliver from 2,000,000 to 10,000,000 gallons per acre per day, and the bed is deeper and of finer sand. They use no coagulant, and are cleaned by scraping at infrequent intervals. The rapid gravity filters will deliver from 100,000,000 to 150,000,000 gallons per acre per day, and their beds are shallower and coarser; they use a coagulant and are cleaned frequently by reverse washing.

After a slow sand filter 'has been in use for a short time, various substances, bacteria, etc., accumulate on the surface, forming a sticky coating that lessens the spaces, or voids, between the grains of sand, and thus helps to retain the impurities that are suspended in the water. Some chemical action also takes place (oxidation, ele.), and this aids the process of purification. In a rapid or mechanical filter, however, the water passes through the bed so quickly that such a coating has little time to form out of natural materials, and certain coagulants are therefore added to produce an artificial coating. A coagulant often used is sulphate of aluminum, which in alkaline waters soon produces a fiocculent substance that permeates the water and afterwards the *368bed, and detains much of the suspended impurity. This flocculent matter is sticky and reduces the pore-spaces between the grains of sand. These rapid filters need cleaning frequently (as often as once a day, or even oftener), and this is accomplished by draining off the coagulated water down to a certain level, and then injecting from below a stream of clean water. This water disturbs the sand thoroughly and carries off the impurities that have accumulated in the bed. The process of washing by an upward current is usually assisted by mechanical agitation, as by arms or rakes, or by forcing a blast of air up through the bed.

[2] Several matters should be briefly alluded to.'*' “Positive head” means the weight of the water both within and above the sand bed, and is usually measured from the controller or the underdrainage system to the surface of the water to be filtered. “Negative head” is the force that comes into play when a partial vacuum is created either within or below the filter bed. Under usual conditions, negative head cannot begin to operate until the bed becomes somewhat clogged with material. A gravity filter may operate exclusively by positive head, and all gravity filters use such a head to some extent; but a negative head filter uses both kinds of head, and is therefore provided with some means for producing a partial vacuum within the bed. This may be a pump, a trapped pipe, or some equivalent device. The trapped pipe may drop below the bed, thus adding to the total available head; or it may lead away horizontally at the bottom of the bed.

[3] Sand of “effective size” means sand containing 10 per cent, by weight that is finer in grain than the diameter sought to be secured. The ideal sand would have all its grains of uniform size, but, as this cannot be attained, the size is “effective,” if about 10 per cent, by weight is finer. It is also true that some percentage of the grains is larger than the particular diameter aimed at. In all municipal filtration plants, careful provision is made for preliminary treatment of the water before actual filtering begins. This consists either of sedimentation or coagulation, or of both, and thus a certain quantity of suspended impurities is removed, and the filters proper are relieved from much of the work they would otherwise have to do.

In the specification accompanying the apparatus patent of 1895, and also in the specification accompanying the reissue, the invention involved in the present suit is described generally as follows:

“This invention relates to improvements in ‘gravity-filters’ of the kind adapted for filtering large masses of water, and which are provided with a power-driven agitator for stirring up the filtering material during the process of washing and cleansing the same. The object of the invention is to provide a filter having the highest efficiency, and which can be constructed and operated at minimum cost, for material and labor. The invention consists, briefly, in combining a filter and subsiding-chamber in one tank or vessel occupying but one foundation, and so constructed and arranged relatively that the accumulation of sediment in the subsiding-chamber may be conveniently removed and driven out therefrom by mechanical appliance and without other power or machinery than that found in connection with the agitator, which is thus utilized for this purpose. The invention further consists in the novel features of construction, combination, and arrangement of parts hereinafter described and set forth; reference being had to the accompanying drawings."

*369A description of the device is then given, in which a filtered-water pipe leading to the reservoir outside the tank is referred to. The specification then goes on:

“ * * * Said pipe is provided with a water-trap 9. Said filtered-water pipe is extended some distance below the filter-tank to form a downdraft by removing the air-pressure from the under side of the filter-bed, and thus to make effective the air-pressure upon the water above said bed, and said trap is for the purpose of sealing the lower end of said filtered-water pipe to prevent the air from entering therein above said trap when the flow and velocity of water down the pipe is diminished by the accumulation of impurities in the bed.”

In the paragraph just quoted appears 'the only reference to the presence and effect of a partial vacuum in the filter-bed; the vacuum being caused by the trap in the downdraft pipe. The single claim of the original patent referred to the subject (if at all) only by the implication of the italicized phrase:

(1) “In a filter the combination of a filter tank, a subsiding tank, and a filter material holding tank, formed as described, with a central communicating passage or opening between said tanks, an agitator for said filter, the shaft of which passes through said central opening, and projects Into said subsiding chamber, and a pipe elbow rigidly secured to said shaft, and adapted to be rotated thereby, to give force and direction to the water passing down through said central opening and elbow, for flushing out and removing the sediment from said subsiding chamber, substantially as set forth.”

But, when the reissue was applied for, this subject of partial vacuum was so expanded and emphasized as to become much the most important, as will appear from the following quotation of new matter in the 'specification:

“As_ soon as the water passes into the filter-tank from the subsiding-tank it begins to percolate through the sand or other granular material composing the filter-bed, and after passing through the bed is carried ofl: by the filtered-water pipe 8. When a continuous flow of water is secured through the filtered-water pipe 8, the downdraft exerted by the column of water In said pipe creates and maintains a partial vacuum in the filter-bed; the vacuum being greatest in the lower portion of the bed. The granular filtering material is consequently compacted by atmospheric pressure, the lower part of the bed being compacted more closely than the upper part; the degree of compactness gradually diminishing toward the upper surface of the bed. As the result of this condition of the filter-bed, the impurities carried by the water to be filtered are not all collected at the surface of the bed, forming a crust or substantially impervious stratum, as is the case where granular filter-beds of substantially uniform density are used without suction, but only a part of the impurities are removed at the surface; a great portion of the suspended matter, by reason of its being in a more finely-divided state, being permitted to pass tile more widely-separated granular particles in the upper portion of the bed to bo intercepted afterward by the more closely-eompactcd particles in the lower portion of the bed. Furthermore, by employing the downdraft, as above described, the upper portion of the filter-bed, which usually contains a comparatively large proportion of the impurities, may be agitated by means of the agitators without interrupting the process of filtration, inasmuch as the lower portion of the bed is maintained in a compact state by the downdraft, preventing the passage of the impurities freed by the agitation of the upper portion of the bed, the partial vacuum maintained by the downdraft acting to prevent the agitation of any part of the bed except that which is directly operated upon by the agitators.

*370“In prior filters, where a loose granular filter-bed was used, it was impracticable to agitate the filter-bed during filtration, because, as soon as the upper crust of the bed was broken by the agitation, the Impurities would at once be permitted to pass entirely through the bed and be carried off in the filtered water. In my improved filter the agitation of the bed may be repeated as often as may be desired, or it may be continuous, thereby permitting the mass of the impurities to penetrate further into the bed until the entire bed becomes saturated with impurities, when it is necessary to wash the bed, as will be hereinafter described.

“It will be evident from the above description that a filter constructed, as herein described, is capable of use without washing for a much longer period of time than prior constructions, thereby effecting a great saving both in labor and in the water used for washing. Furthermore, the capacity of the filter is greatly increased, ás by distributing the impurities more or less uniformly throughout the bed, and preventing the formation of a thick crust at the upper surface, the water is caused to pass much more rapidly through the bed without interfering with successful filtration.

“The maintaining of a partial vacuum in the filter-bed, as above described, further improves the operation of the filter and increases its efficiency by effecting the coagulation of the suspended impurities; such coagulation being effected by the separation and concentration of the air contained in the water under filtration, thus increasing the bulk of such impurities and enabling the filter-bed to intercept and retain them.* The air separated from the water remains to a great extent in the filter-bed¡ and is especially useful in washing the bed, as when the current of water is reversed in washing, as hereinafter described, the air rising through the filter-bed causes a violent agitation thereof, liberating the accumulated impurities and permitting them to be carried up with the wash-water to the overflow. (In filters of this class daily washing is usually required, while the sediment in the subsiding-ehamber may accumulate for a month or more before its removal is necessary. The daily wash is performed by closing the valve to the subsiding-chamber, filtering the water down, closing the valve in the pure-water pipe, and opening connection therewith to the wash-water, stand-pipe, or pump, forcing the water up through the filter-bed, during which the agitator is kept in motion and the washed-out impurities flow over the top of the material-bank into the annular space surrounding it and pass out through the valve-opening at the bottom thereof. The wash-water is prevented from getting into the subsiding-chamber by the central pipe, which projects above the top of the material-tank for this purpose. The wash is completed by closing the valve of the annular space and permitting the first filtered water to pass out into the sewer instead of the pure-water reservoir.) As soon as the downdraft again becomes effective, the filter-bed is at once compacted to such an extent that it is necessary to draw off only a very small amount of water after washing.”

The part we have placed in parenthesis appears also in the original specification, but the rest is new. The theory that was thus elaborated in the reissue is further dwelt upon in the specification of the method patent:

“My invention relates to purifying waters for potable purposes by filtering such waters through granular filters. As is well known, the principal object of filtering the water supply of towns and cities is to remove the suspended impurities, which in many instances are in a finely-divided state, and experience has proven that granular filters (that is to say, filters in which the water is caused to percolate through a filter-bed composed of loose sand or other granular material) furnish a filtering medium which is best adapted for removing'' the suspended impurities while permitting the water to flow at a rapid rate, as is necessary where the water supply of a town or city is being filtered. In using granular filters heretofore, however, it has been found that in most instances a granular filter-bed will not intercept and retain the suspended matter in the normal condition, and it has therefore been found necessary in almost all instances to introduce a coagulant, such as alum, into the *371water before filtration, the coagulant serving to coagulate the suspended matter into masses which are larger and more susceptible of interception and retention by the granular particles composing the ’filter-bed; in some instances it being necessary to use such a large quantity of alum as to affect the taste of the filtered water.

“My present invention consists of an improved process of purifying water by filtration through a granular filter-bed by which the necessity of the use of alum or other coagulant is in many instances entirely avoided, and in other instances, where the impurities are in an exceptionally finely-divided state, the quantity of alum which under former processes would be necessary is very greatly reduced.

“To this end my invention consists in effecting what may be termed the ‘coagulation’ of the suspended impurities of the water by suction while passing through the filter-bed, the particles of suspended matter being thereby caused to come together into masses of sufficient size and of such character as to be readily intercepted and retained by the granules composing the filter-bed.

“My invention further consists in applying the suction principally at the lower portion of the filter-bed, so that it acts more strongly upon the finer particles of suspended matter which have passed through the upper portion of the filter-bed.

“My invention further includes the compacting of the filter-bed in such manner that the lower portion thereof will be of the greatest density, the density gradually decreasing toward its upper surface, as by this means, while the larger masses of impurities will be retained by the more widely separated granules at the upper portion of the bed, the lower portion of the bed will be sufficiently dense and compact to intercept the smaller particles of suspended matter, especially after they have been coagulated, as above stated.

“In the accompanying drawings I have shown apparatus designed to utilize my improved process, each apparatus in general consisting of a filter-bed of loose granular material contained in a suitable receptacle and a pure-wator pipe in the bottom of said receptacle, said pipe being provided with suitable strainers to prevent the granular material from entering such pipe, and with an off-carrying pipe vertically arranged and of such length that, as the filtered water is carried off by said pipe, a partial vacuum will be created within the filter-bed; the vacuum being greatest in the lower portion of the bed and gradually diminishing toward the upper surface thereof. By this means, a continuous suction is exerted which is greatest in the lower portion of the bed, compacting it so that its density will be greatest at the bottom and will gradually diminish toward the upper surface thereof. Furthermore, the suction causes the air contained in the water to separate and concentrates it; the minute bubbles coming together, forming larger bodies of air, which are to a great extent retained within the filter-hod. As the fine air-bubbles converge they act to concentrate and coagulate the particles of suspended matter, so that finally the suspended matter is formed into larger bodies, which, may easily be intercepted and retained by the filtering material. As the process of filtration continues, the air extracted from the water gradually accumulates in the bed, still further compacting it and increasing the efficiency of the filter to such an extent that, even though the bed contain large quantities of impure matter extracted from the water, the filter may nevertheless be continued in use with satisfactory results, thus making it unnecessary to wash the bed as frequently as has been necessary with other forms of filters employing granular filter-beds. * ’* *

“Referring to Fig. 1, which, as above stated, illustrates the condition of the granular material after filtration has commenced in accordance with my new process, it will be noted that the granular matter in the lower portion of the bed is compacted to a much greater extent than that at the upper portion thereof; the degree of compactness gradually diminishing as the surface of the bed is approached. Such compacting of the bed is effected by the creation of a partial vacuum, principally in the lower portion of the bed, by means of the downdraft through the pipe 10. By thus compacting the bed it is much better adapted for removing the finer particles of suspended matter than would be the case were the bed of uniform density. Furthermore, the-*372entire body of filtering material is, utilized instead of substantially tbe upper surface only, as in prior constructions.

“In Fig. 2 it will be noted that in tbe lower portion of the bed there are a number of air-spaces; the larger air-spaces being at the lower portion of the bed. Such spaces are formed in the bed by the accumulation of the air extracted from the water by suction, as hereinbefore described. After the air has performed its office of coagulating the suspended matter, it accumulates Within the bed, as illustrated, and afterward is especially useful in washing the bed, as, when the current of water is reversed for the purpose of washing, the air being liberated rises through the bed and, escaping, agitates it violently, thereby freeing all the accumulated impurities, so that they may be carried off with the wash-water.

“The manner in which the impurities are coagulated by the extraction and concentration of the air contained in the water is clearly illustrated in Figs. 3, 4, 5, and 6. Fig. 3 illustrates, on a magnified scale, the condition of the air. and the minute suspended impurities in the water; the circles representing the air-bubbles and the black marks the suspended impurities. It will be noted that the air-bubbles, as well as the impurities, are at first comparatively widely separated from each other. After suction has been applied to the filterrbed for a short time, the air-bubbles are drawn together, carrying with them the suspended matter, as shown in Figs. 4 and 5, until, as shown in Fig. 6, the air-bubbles finally come together, forming larger bubbles, the suspended impurities also coming together, forming clots or masses of sufficient size to be intercepted and retained by the granules of the filter-bed. In this way, with many waters, practically all the suspended impurities will be coagulated and removed without the use of a chemical coagulant, while with those waters in which coagulants must be used the quantity of the coagulant which it is necessary to use is greatly reduced.

“While in the apparatus shown in the drawings the necessary vacuum is secured by the action of the pure-water discharge, I do not limit myself to securing the vacuum in that way, as other means may be employed to secure such result. I prefer the apparatus shown, however, as by it the suction is applied uniformly through the lower portion of the bed and is substantially uniform in any given stratum.”

Then follows a disclaimer in view of the prior art:

“I am aware that heretofore atmospheric pressure has been employed to force water through filtering material, such as stone or felt, but such use has been simply for the purpose of forcing the water through the filtering substance, and has not effected the extraction and concentration in the filtering material of all or a part of the air contained in the water or the coagulation or agglomeration of the impurities, as described; nor has it secured a bed of filtering material of varying density or compactness. My invention, therefore, does not contemplate the use of filtering materials such as those above referred to, but is limited to the use of'a granular filtering material having an exposed filtering-surface, by which I mean a surface uncovered by felt or similar material which would prevent its use in connection with my improved process.”

The basis of the decision of the District Court was that in Jewell’s patent a pull of suction exists whereby a pulling force is created upon the material in the bottom of the filter, and that the use of such suction action tends to release air from the water which may be of advantage, and also tends to cause the sand grains in the vicinity of the strainer system to reseat themselves, and thereby compacts the section of the bed. In his opinion, among other things, the learned judge said :

“ * * * That the process and apparatus disclosed by Jewell * * * do in practical use, by the creation of a vacuum, utilize the whole sand body *373as a filtering agency, as it was never used before. Indeed that the workings of a downdraft or negative filter must in the nature of things be different from a gravity or positive head filter is self-evident, from the fact of its calling into play the powerful agency of atmospheric pressure. From this we can readily understand that such atmospheric force, whether venting itself as a weight from above or a suction from below, must cause new results in a process, the basic element of which is the pressure passage of liquid through impeding matter.”

Thus the reasoning of the court was in effect that the process and theories disclosed by Jewell in his patent impractical use did, by the creation of a vacuum, utilize the whole sand body as a filtering agency as it had not been used before, and that the Jewell process does by the use of a downdraft, sealed, vertical outlet pipe create and maintain an operative vacuum, which vacuum effects a deeper utilization of the sand body for filtration than in positive head filters; that the presence of a released air incident to the use of a downdraft vacuum is helpful and not harmful by reason of the velocity imparted to the passing water by such vacuum; that the process makes the runs longer; and that both structural and maintenance costs are lessened by its use. Now it appears that, in the art of filtration, it has been the practice, when it is desired to lengthen the run of a filter or to obtain the more rapid passage of the liquid through the filtering material, to utilize atmospheric pressure upon the liquid by creating a partial vacuum within or under the filter, in this way unbalancing the atmospheric pressure which had previously been equal on both sides of the liquid. The way of accomplishing this is called “suction,” and in the language of the patents in suit is described as “suction from below.”

Application of this way has been accomplished in filters of many kinds. In British patent, No. 4393, of 1819, issued to Tritton for a filterer, the process was described to be one by means of an air pump or other mode of producing exhaustion or vacuum in the part or parts of the apparatus into which the filtered liquid flows, and thereby to create a difference between the atmospheric pressure acting on the liquid before and after filtration, and by means of that difference of atmospheric pressure more actively to force the unfiltered liquid through the pores, apertures, and interstices of the filter. Tritton also describes in his apparatus a part of the filter intended to receive the unfiltered liquid as communicating with the external air in such a manner that, when such part of the filter as is intended to receive the filtered liquid and the receiver are exhausted by an air pump or otherwise, the power of atmospheric pressure on the surface of the unfiltered liquid, together with the pressure used in ordinary filtration, may force it through the filter or filters by which it is to be strained.

In British patent, No. 6160, issued to James Neville (1831), and cited by the examiner in all of Jewell’s applications, the invention being entitled “An improved apparatus for clarifying water and other fluids,” the patent referred to a downdraft tube trapped, which, when filled with any fluid of the specific gravity of water and the cock is turned, the action of the atmosphere on the surface of the fluid contained in the cistern will be with immense force, whereby a large *374quantity of the fluid would be driven through a clarifying medium in a short time. Neville used a filter bed of granular material and described his apparatus, stating that, in carrying the pipe as low as possible, he did so in order to obtain greater atmospheric pressure on the surface of the fluid containéd in the vessel or reservoir, by which means a greater quantity of water would be forced though the clarifying medium. Neville disclaimed novelty as part of his invention, for the use of charcoal or sand as a clarifying medium, because that had been known and acted on for a long time before.

So, without going into greater detail, it is clear that, for many years before Jewell went into the field, there were filters with granular filter beds having exposed filtering surfaces with means for producing a partial vacuum in the filter bed and the actual application of suction from below, as well as means for reversing the flow of water through the filter.

In British patent, No. 8371, of 1899, to Edwards for “improvements in the prevention of the pollution of river waters and water courses from manufacturer’s polluted waters and waste, and for cleaning and purify ordinary river waters,” the filtering materials used were burnt coke reduced to small proportions, well washed river sand, and all or any or either of these materials, as best suited the special conditions under which the filter might be operated. The patent, after describing a siphon pipe and an ejector and a pipe for supplying steam, said that, in order to draw off the filtered water from each subcompartment, the inventor introduces into it one limb of a siphon, to which he attaches a steam jet ejector, by which he was enabled to acquire any required vacuum in the subcompartment and below the filter-bed. His patent also stated that such vacuum, if found to be more than required for perfect filtration, could be regulated by means of a small tap in each compartment to admit a portion of air when necessary.

The .Lawrence city filtration plant, a slow sand filter, disclosed that in 1892, and prior thereto, trapped downdraft pipes were used upon slow sand filters having exposed filtering surfaces. They were apparently used in 1893 in the Lawrence city filtration plant, and in Hazen’s book on the Filtration of Water Supplies, written before April, 1895, the author described the Lawrence experiment station filters.

In the Crocker mill filters, installed in 1891 at Holyoke, Mass., there were filter tanks containing filter-beds of sand with exposed filtering surfaces. In these filters, it is evident, from the testimony, a partial vacuum was created and maintained in the same sense as in the patents in 'suit.

In August, 1895, there was published in the Paper World the description of the Moore filter; the article referring to high and low pressure filters, and describing high-pressure filters as distinguished from the low. It was there said that, in the low-pressure class, a strong wooden case,-open at the top, was employed, the water flowing as required, and after percolating through, the sand was conducted away; the latter operation being generally aided by a pump, which, *375establishing a partial vacuum, aided the passage of the water by suction. This reference appears to have been made as if such a use has been a common one, and certainly it fits a description of the operation of the filters of which the Crocker and one other filter theretofore in operation furnish instances. Prior art also shows means of producing a partial vacuum or suction in a filter, as illustrated by patents, such as the Neville, O. H. Jewell, No. 423,430 (1890), Sinclaire, No. 171,056 (1875) and Simmons No. 59,467 (1867). -

The principle of physics which controls is that by' diminishing the atmospheric pressure below (that is, by unbalancing the pressure), the atmospheric pressure above, which remains constant, always becomes effective for pushing or forcing the liquid through the filtering material. It therefore seems certain that nothing new can be claimed for “suction,” or the creation of a partial vacuum within or beneath a filter-bed. Again, the compacting of the filter-bed greater at the bottom was described in the patent issued to Hyatt, No. 293,745, and, as a means employed to produce “suction,” there are instances in which a trapped downdraft pipe has been used. In the present case, the patentee is confined to a downdraft pipe, trapped, inasmuch as untrapped downdraft pipes were well known in the art before the invention claimed by the patentee herein. The patents we have examined as introduced in evidence show that there were in use granular beds having exposed filtering surfaces in combination with suction, and that, when Jewell in 1895 made application for his original patent, there were means very like his for making effective the air pressure above the filter-bed. Again, the reissued patent uses the term “suction,” based upon the idea that there is a pulling effect so compacting the lower portion of the filter-bed by the presence of a partial vacuum beneath it; and in the method patent the terms “suction” and “suction from below” are used with an attribution to them of a drawing effect. A witness for the filtration company, after describing compacting action under suction, states as follows:

“Third. Under these conditions the weight of the column of water descending in the suction or downdraft pipe causes the production of a larger output of properly filtered water; and the sediment separated from the water during filtration is carried farther down into the bed by the weight and velocity of the descending column of water below the bed, which force acts, as the upper portion of the bed becomes choked with sedunent, to draw portions of such sediment layer downwardly into the bed, the upper portion of the bed being maintained in looser condition than it is ip. a positive head filter, and so allowing the free entrance of the suspended matters, and the lower portion' of the bed being more compacted than the upper part, as before stated, and so tending to retain such impurities within the bed.”

The “pulling of suction” theory does not stand under critical examination, however, for the testimony of scientific witnesses proves that the operative force of what is commonly called “suction” is not a pulling, but is atmospheric pressure exerted upon the liquid or other movable material to the extent of the difference of pressure upon the opposite sides of the movable material; that is to say, the operative force is unbalanced atmospheric pressure, and this atmospheric pressure acts in the direction of the area of diminished pressure, so it *376becomes a pushing force, well described by one of the witnesses for the appellant in the following simple language:

“The operation of a partial vacuum in the lower part, or any other part of a filter, simply means that at a section above that part of the filter the sand is so clogged up that the water cannot pass through it as fast as it can run out of the outlet valve, and there is a tendency for the water in the filter to pull apart and leave in the sand a space not occupied by water (a partial vacuum), into which the air pressing on the surface of the water on top of the filter drives the water 'with increased force, measured by the degree of vacuum produced. Therefore, when a partial vacuum exists in a filter-bed, there is an added force applied on the surface of the water in the filter; tending to drive the water through the sand; under no circumstances can there be a force applied from below to pull the water through. The application of the force is from above downward, and acts, in driving the water through the sand, precisely as would an added equivalent'head of water.”

Another expert witness, whose scientific attainments are very clearly established, said:

“Suction does not mean that there i? a pulling of the water: in fact, water has not cohesive properties to permit it to be pulled, but it must move either by being pushed or else by the force of gravity. Hence the expression ‘suction applied from below,’ intimating that there is a pulling force upon the bottom of the filter, refers to something that does not exist, and a force is actually applied at the top of the body of water resting upon the sand bed in the filter by virtue of the unbalancing of the atmospheric pressure.”

It is proven that a partial vacuum is not made when the filter begins operation; nor is it shown that such vacuum is always greatest at the bottom of the bed. The supply from the filter-bed being regulated by a device or controller, the effluent pipe delivers less than the full capacity of the filter. From these facts it follows that, if we have a bed of clean sand, there can be no unbalancing of atmospheric pressure so long as the head presses the water- through the bed at the rate set by the regulating device, and that it will be only when a clog happens that the effluent pipe will carry off water quicker than it is delivered and that a vacuum can occur. As explained by one of the witnesses for the appellant, the statement in the patents that the vacuum will be greatest at the bottom of the filter-bed is erroneous, if it is understood that such vacuum is normally and at all times or even usually found at the bottom, because such partial vacuum is a variable phenomenon conditidned upon the interval which has elapsed since the filter may have been washed, the degree of fineness of the superficial layer of sand, as compared with that below it, and the degree of clogging of the filter. The clogging would seem to be likely to be anywhere, depending upon size and kind of sand, rate of flow, presence of air, or other factors; hence the partial vacuum may occur at any time or place apparently not capable of determination by device or method of the patent.

We are uncertain whether or not, as a result of a partial vacuum at the bottom of the bed, there is a greater compactness at the bottom than above, or, as it is said, whether there is a differential compacting from below upward. Prof. John W. Langley, a distinguished physicist, said that such differential compacting did not occur in any filter where cleansing by upward washing was a feature of the action. He *377said that tlie upward wash brings the finer grains of any heavy granu- ' lar material to the top, forming thus a subsidence layer of such material of variable depth, in which the compacting is greater than in any other part of the bed. He added that any differential compacting which might occur below this top fine sand stratum would be very small in amount, and would be the result of natural forces, whereby a very small additional compacting of the sand might be near the bottom, due possibly to the friction of the water passing down through the sand as a necessary consequence of the properties of water and sand. Dr. Mason, another distinguished scientist, said that it would be impossible to make any actual measurements as to compacting of the filter-bed or near the strainer openings within the filter here involved, so as to determine the matter experimentally, and that, while he believed there would be an action toward compacting in the lower part of the bed, he could not say how far such action would extend, or what its amount would be, or whether it would be progressively smaller in a uniform degree.

We might quote at length from what scientific men testified to, but it would make our opinion needlessly long. We are agreed, however, that, when we weigh the evidence of one side against the other, it is highly probable that the lower part of the bed is but little affected by the so-called “pull of suction,” and that there is but very little compacting by compression of the grains of sand, or by the sand grains moving because of the superincumbent weight of water and air. Moreover, such compacting of the lower part of the bed as occurs under the weight of water and air is not peculiar to filters operating under negative 'head. Neither has it been proved that the friction of the filtering water compacts the bed by rearranging the sand grains throughout the bed; and, in any event, such rearrangements would be as likely to occur in a filter under positive head as in a filter under negative head. There may be some rearrangement of sand grains near the strainer system; but, if so, it lacks the support of experiment, and would seem to be inconsiderable in extent, and as likely to be present in a filter under positive head as in a filter under negative head. Moreover, we think it is true that compacting due to overlying weight or to the downward flow of water is not the compacting contemplated by the patents. And, after a mechanical filter has been washed by a reverse current of water and air, the sand is finest and most compact at the top, because the bed must then rearrange itself in accordance with the subsiding vaittes of the grains.

With regard to the assertion that the suspended impurities penetrate the bed more deeply under the operation of the patents, the only direct testimony on the subject is unsatisfactory. Some of the testimony relates to observations made from 8 to 15 years before the witnesses were examined; and, as no records were kept, we regard it as unsafe to rely on the unsupported recollection of witnesses after the lapse of so long a time. One other witness made a single experiment under conditions that were not very fully described and were apparently not conducted with caution. The other testimony on this point on behalf of the plaintiff was largely theoretical, and is so colored by the under*378lying assumption that “suction” is a pulling force as to be much impaired in value. Indeed, misconception of the principle of “suction” has destroyed appellee's case, for it is the real foundation thereof.

With regard to the liberation of air in the bed and its usefulness there as a “coagulant,” the weight of the evidence seems to indicate that air does not of itself act as a coagulant, and that its presence in a filter-bed is almost always objectionable.

The objections to the theories of the patents have been strongly put m the able brief for the city, and we can do no better than quote them as follows:

“ (1) In accordance with the pretensions of the patents, the whole invention of Jewell resides in the compacting of the granular filter-bed, greatest below and gradually diminishing towards the top, and the extraction and retention within the filter-bed of a greater or less portion of the air contained in the' water. The cause of such differential compacting is alleged to be the formation of a partial vacuum in the filter-bed, greatest below and diminishing towards the top, which partial vacuum is also claimed as the cause of the liberation of air. The beneficial results of such connpacting are claimed to be debper penetration of suspended impurities into the filter-bed, and the consequent greater utilization of the entire filter-bed, and, of the liberation of air, the agglomeration or coagulation of suspended impurities, so as to dispense, entirely, or to a great extent, with the use of a chemical coagulant, and the greater agitation of the filter sand in washing by a reverse current. Except the mere fact that air is liberated from water by diminution of pressure or the formation of a partial vacuum, every one of the pretensions is controverted and has, we believe, been shown to be false.

“ (2) In a so-called negative head filter, as ordinarily constructed and operated, a partial vacuum does not occur at the commencement of the run and continue throughout it. It may not occur at all and, if at all, will occur only after the conditions of clogging have become such that the positive head, or pressure due to the superincumbent water, has been utilized in forcing the water through the clogged area or plane of greatest resistance, and this will be towards the end of th'e run. In the Harrisburg filters (said to be infringements), it occurred at all in only 25 per cent, of the runs, and then only for a short time at the end of the run.

“ (3) Actual observation shows that it will occur, not necessarily or usually first or greatest at the bottom of the filter-bed, but always immediately below the plane of greatest resistance, which may be anywhere in the filter, from the strainer system up. In the Harrisburg ■ filters it occurred indifferently at the top or bottom of the filter-bed or anywhere between, and was ‘greatest in the lower portion of the filter-bed,’ 115 minutes in nearly 5 months, or .06 per cent, of the entire time the filters were in operation.

“(4) The time and place of the occurrence of a partial vacuum depends upon various conditions, viz., amount and character of suspended impurities, size of sand, rate of flow, use of coagulant, temperature of water, eta, and are wholly beyond the control of the operator. Complainant's testimony is of little or no weight, because it fails to take these things into account

“ (5) In a negative head filter there is no compacting of the filter-bed, greatest below, due to the creation and maintenance of a partial vacuum greatest in the lower portion of the bed, for the reasons: (a) That such vacuum does not so appear usually, but only occasionally, and then only at the end of a run; and (b) that such vacuum, when it occurs, is incapable of producing such compacting. Neither is there such compacting due to the pulling or drawing of ‘suction’ upon the sand grains, for the reason that what is commonly called ‘suction’ has no such effect, but operates merely by rendering effective the atmospheric pressure upon the water column within and above the sand bed, and thus adding to the effective head in the same manner as an additional depth of water or any other additional pressure.

“ (6) Such differential compacting of the lower part of the filter-bed as may *379be due to weight of superincumbent sand and water, by way of either compression or movement of the sand grains, is negligible in amount and has no effect upon the operation of the filter. It has no relation to suction and will occur, if at all. as well in positive as in negative head filters.

“(7) Movement of sand grains due to velocity of flow is, as to the lower part of the filter, also negligible in amount. It will occur more in the top layers of the sand, where the grains are free to move, than in the lower portion of the bed, where they are held in a state of great fixity by the superincumbent weight. It has no relation to suction, and there being no difference between negative and positive head filters, based upon rate of flow, will occur in tlie same manner and to the same extent in the latter as in the former.

“(8) Differential compacting due to superincumbent weight or velocity of flow is not the kind of compacting called for by the specifications and claims of the patents in suit, which is compacting alleged to he due to underlying vacuum or ‘suction from below.’

“(0) In all filters whose sand beds are washed by a reverse current of water (which is the universal practice in all rapid or mechanical filters), compacting is always greatest at the top of the bed, by reason of the subsidence, after washing, of the finer grains of sand at the top and the coarser grains below. This is practically an admitted fact in this case, being testified to by defendant’s four expert witnesses and substantially by complainant’s witnesses, Dr. Mason and Mr. Milligan, and is not denied by any of complainant’s witnesses, although there was ample opportunity to do so; and no attempt was made to show that such compacting, greatest at the top, is overbalanced or offset by any compacting below.

“ (10) Penetration of suspended impurities is a function of the flow of the water carrying them. The depth to which the impurities will penetrate into the filter-bed depends upon amount and character of suspended matter, size of sand, rate of flow, amount of coagulant, temperature of water, etc., and is not affected by the character of the actuating head, which, whether positive or negative, is, as we have seen, pressure — in the one case, hydrostatic, and in the other, hydrostatic and pneumatic.

“(11) Penetration of the filter-bed is not effected by partial vacuum greatest in the lower portion of the bed, because, if the latter occur there at all, it is oniy occasional and of brief duration, and because it is incapable of producing such penetration; nor is it effected by the ‘pull of suction,’ for reasons already stated.

“ (12) Utilization of the entire filter-bed and deep penetration of suspended impurities is neither desirable nor safe,i because of the danger of the impurities passing entirely through the filter-bed and fouling the filtered water. It is always desirable to reserve the lower part of the filter-bed as a factor of safety to meet any abnormal conditions that might develop, and thus prevent deterioration of the effluent.

“(13) The possibility of deep penetration with safety to the effluent would depend upon the compacting of the filter-bed greatest in the lower portion. This is clearly admitted in the patents and is necessarily true. But such compacting does not occur in negative head filters using a reverse current in washing, for reasons already stated.

“(14) The liberation of air in the filter-bed, far from being useful, is a nuisance, and the greatest objection to negative head filters, because it causes ‘air binding’ of the sand bed, and because, when the air-bubbles grow large enough, they rise by their buoyancy and ‘break through' the sediment layer, causing channel-ways and irregularities of filtration which seriously affect the effluent. 1

“(35) It is shown that in the best modern filter practice liberated air is gotten rid of as far as possible, and complainant has attempted to show that the velocity of the moving water carries the air off through the effluent pipe. But, if it is so carried off, it cannot perform the useful functions claimed for it in the patents. Besides, if it is carried off below, it clogs the drainage pipes, reducing their capacity and causing unequal rates of filtration when large bubbles of air are suddenly carried out. ‘Whichever way it goes it makes trouble, decreasing the efficiency of filtration or the length of the run, and frequently both.’

*380“(16) Liberated air is not useful in washing the filters. This pretension was rejected by Dr. Mason and by defendant’s witnesses, and was disclaimed by complainant’s counsel at the oral argument.

“(17) The pretense that air bubbles are useful in agglomerating or coagulating suspended impurities is also without foundation. It is rejected by Dr. Mason, as well as by defendant’s witnesses.

“ (18) The same is to be said of reduction in the amount of coagulant used.

“(19) The liberation of air depends upon the reduction of atmospheric pressure or creation of a tendency to a vacuum, and hence depends upon the conditions above stated relative to the formation of a partial vacuum, and time, place, and amount are wholly beyond the control of the operator. It being an admitted fact that liberated air, if not gotten rid of or controlled, is harmful, the patents, while claiming its utility, fail to point out how it can be controlled so as to be made useful and not harmful.

“ (20) All the above-mentioned things, maintenance of a vacuum greatest in the lower part of the filter-bed, compacting of the bed greatest below, deeper penetration of suspended impurities, utilization of entire filter-bed, coagulation of impurities by air-bubbles, and other utility of liberated air, upon which the reissue was asked for and the method distinguished from that of Neville, and upon which both patents are built up, are either admitted or shown to be without existence. All that remains, as the only useful function of the Jewell apparatus or effect of the Jewell method, is the unbalancing of atmospheric pressure so that the atmospheric pressure above the filter-bed may become effective to force the water through the filter-bed when the same becomes clogged; and this is expressly disclaimed in the method patent, because Neville had already done it, and is what many others had been doing for 75 years before Jewell came into the field.”

In the main, these criticisms commend themselves to us as well founded. However, it is probably enough to say that (at the best) the evidence in behalf of the company leaves us in much uncertainty whether the theories of the patents are sound. We do not feel bound to go the length of declaring the patents invalid; the reissue patent,"indeed, has already expired, leaving the patentee only a claim for royalties; but we are fully prepared to say that, while the subject is so surrounded by uncertainty, the company cannot reasonably object to'the application of a test that tries the Harrisburg filters by the theories of the patents. If these filters do not produce the results called for by these theories, it is certainly fair to conclude that the company has not succeeded in making out the charge of infringement. On this point, there is one piece of evidence that is not referred to by the district court, but in our opinion is of decisive weight upon the issue of infringement. We mean the prolonged experiment that was made with the Harrisburg filters, the results of which may be summarized as follows: Filter No. 2, which is typical of the 12 filters operated by the city, was prepared for the experiment under Prof. Fuertes before January 9, 1909, when a preliminary or trial run was made. It was provided with a series of gauges to show loss of head, and these indicated upon a scale the pressure at seven levels — the first in the water just above the filter bed, the next four at different levels within the bed of sand, the sixth in the gravel, and the seventh in the outflow pipe carrying the filtered water away. These gauges were read hourly from January 21st to June 11th, inclusive. Diagrams, together with the record of these readings, were given in evidence; 256 runs of the filter were covered during the winter, the spring, and the early *381summer, when the water varied in temperature from 36° to 74° Fahr., and when the turbidity varied between 5 and 85 parts in a million gallons. There could hardly be better evidence of what goes on within an exposed filter-bed of sand, where means exist for producing suction, so far at least as a reduced pressure or a tendency to a vacuum is concerned. The filter was operated in the usual manner, so that nothing exceptional impaired the value of the experiment. Now, when the assertions of the patents are remembered, namely, that as the water flows down the trapped pipe a partial vacuum is created and maintained, that the vacuum is greatest below and gradually decreases towards the top, and that the filter bed is thus compacted, more densely below and gradually less towards the top, it is significant that the records of filter No. 2 do not bear these assertions out. On the contrary, they show that a partial vacuum never came into being at or near the beginning of a run; that 191 runs out of 256, or-about 75 per cent, of the whole, did not show a partial vacuum at all; and that while 65 runs, or about 25 per cent., showed a partial vacuum, they did not show it until near the end of the run. Further, of these 65 runs, in which a partial vacuum appeared near the end of the run, it appeared at the bottom in only 8, while in the remaining 57 it appeared in different places, sometimes near the top of the bed, sometimes at varying levels between the bottom and the top. To state the matter from another point of view: During the whole experiment, the records show that a partial vacuum appeared first at the bottom and remained greatest in that part of the bed during only 115 minutes, or for only about .06 of 1 per cent, of the time during which the filter was in operation.

It is not easy to exaggerate the value of these carefully conducted experiments, and we think it within bounds to say that, while the Harrisburg filters are charged with infringement, it is remarkable that their customary operation during six months failed to show the characteristics that are asserted to be inseparable from the apparatus and the method of the patents. No counter experiments were made, and no serious attack was made upon these records, or upon the conclusions they seem to justify. In our opinion, therefore, the case may safely be decided by finding upon the evidence that the city has not infringed either patent.

We therefore reverse the decree, with costs, and direct the District Court to dismiss the bill.