These are the defendant’s exceptions in an action to recover for damage caused by the flooding of *275the basement of the plaintiff’s plant in Athol on the night of September 2-3, 1960. The flooding resulted from the giving way of the frost bottom of a rebuilt Hersey-Sparling water meter installed at the plant by the defendant’s water department on July 14, 1960. An auditor whose findings were not final found that the defendant was negligent. The case was heard in the Superior Court by a judge without jury upon the auditor’s report and other evidence, the judge having reserved decision on the defendant’s motion to strike from the report the finding of negligence and also certain statements which the defendant asserts are only recitals of evidence. After trial the judge, subject to the defendant’s exceptions, denied the motion to strike and refused the defendant’s requests to rule that a finding for the defendant was required. The defendant also excepted to certain rulings in respect of the evidence. The judge made no subsidiary findings but indicated some reliance on the auditor’s report. The “Findings of Fact” read: “After hearing the witnesses, and after a consideration of all the evidence, including the ‘Auditor’s Report’ introduced by the plaintiff, the Court finds for the plaintiff on Counts one (1) and three (3) . . ..”
The frost bottom is designed to give way under the pressure of freezing water and thus to avoid injury to the working parts of the meter. It is a disc the upper part of which is fitted (cast?) into the bottom of a heavy ring casting that forms the base of and is bolted to the upper casing of the meter. The base of the frost bottom, of greater diameter than its upper part, overlaps the exterior of the ring casting. Although the meter with detached frost bottom was in evidence and is before us, neither the exhibit nor the testimony discloses just how the perimeter of the upper part of the disc was engaged with the ring casting that surrounds it.1 It is plain, however, that the design is to have an en*276gagement of abutting surfaces of disc and casting so limited by the extent of the contact or the nature of the engagement or both as to withstand ordinary water pressures but to give way under the pressure exerted by forming ice.
The plaintiff disclaims reliance on the doctrine of res ipsa loquitur or inferred negligence and contends that all the evidence, as well as the subsidiary findings of the auditor alone, warranted the conclusion that the defendant was negligent in supplying a meter with an old used frost bottom incorporated therein.
1. The evidence taken at the trial pointed to the conclusion that the cause of the break was a weakening of the metal of the frost bottom due to graphitization so that it was unable to withstand pressure surges or water hammerings caused by the sudden opening or closing of outlets in the main or in the plant line. Graphitization, according to an expert witness for the defendant, is a change and weakening in the structure of the metal “caused by electrical currents set up by dissimilar components of the metal. ’ ’ It is a “chemically induced” phenomenon which is not detectable by visual examination or by any method employed by municipalities. The presence of electric current would hasten the process. Alternatively, the part could have been weakened by metal fatigue — a mechanical effect, resulting from stresses and strains in the course of use; but the only direct testimony was that testing did not show metal fatigue.
The conclusion was plainly warranted that graphitization was more likely to be present in an old used part than in a new one. The attention of the experts was not directed expressly to an old part that had not been in use, but on this record we think it could also be found that this chemical de*277generation due to electric currents set up by the components of the metal itself could take place over a period of years even if the part was not installed in a water system.
The evidence permitted the inference that the broken frost bottom was made prior to about 1944-1945 at which time the manufacturer changed its specification “for thickness of frost bottoms . . . from .125 to .188 inches.” An expert witness for the plaintiff testified that the thickness of the metal at the bottom of the groove of the broken bottom was .0712 inches; this “represented 57 per cent of the old [pre-1944 specification of] 0.125 . . . [and] 38 per cent of the 0.188 post-1945 . . . [specification] for the frost bottom thickness.” The parties in their briefs have assumed, as did the witness, that the measurements in each case are at the same place so as to be comparable. There was testimony of tolerable variances in thickness from piece to piece.2 But the circumstance that the broken part was only thirty-eight per cent as thick as a part made under the post-1945 specifications tended to suggest that it was made when the specification called for less thickness.
There was testimony that the meter was guaranteed for a pressure of 150 pounds. It could have been concluded that due care called for testing the meter after it was rebuilt, and that the meter was tested at the defendant’s workshop where the pressure was not over 135 pounds “unless there are fluctuations. ’ ’ There was no means of testing for higher pressures. The pressure at the plaintiff’s plant except for surges was not over 135-136 pounds but with a hammer the pressure there could go up to 180 pounds, and hammers there and elsewhere are not unusual.
There were electrical ground wires attached to the water pipes near the meter, one about six inches from the meter “next to the wall” and the other about fifteen feet from it. The employee who installed the meter knew of these wires *278(it could be inferred that he had this knowledge at the time of the installation) and in his view these were “ ‘all right’; . . . you are supposed to have a ground next to the wall as the water service comes in from the street side.”
A witness for the defendant, an expert on municipal water systems and hydraulic engineering, who had advised many communities on their water systems and has “been connected with the Athol Water Department since 1949,” testified that the giving way of meter bottoms for causes other than freezing is rare, and that he had “never run into a condition where a new frost bottom gave out in six weeks from graphitization. ” “ [H]e ha [d] no evidence . . . that the plaintiff caused graphitization in the system; . . . [it] could be hastened by electrical input ... by ground wires but . . . there could be many things which could affect it.”
The plaintiff’s expert testified that the difference “between the .0712 thickness of the broken frost bottom and the .188 or .125 manufacturer’s frost bottom specifications would be a significant factor in bringing about the break.” A witness for the defendant testified that variation from the manufacturer’s specification would have to be over fifty per cent to be important. Even with such a difference a “Hersey meter should be able to withstand a pressure of 450 pounds minimum. ’ ’
A metallurgist, called by the defendant, testified that he had found graphitization in the part, that the presence of an electric current passing through the meter would “tremendously accelerate” the process, and that the presence near the meter of two grounds from electric motors would accelerate it. His opinion of the cause of the failure was “graphitization . . . caused by electrical currents.”
A development engineer in the employ of the manufacturer of the meter testified to the effect that it would be careless not to replace a frost bottom that showed some rust and corrosion and was known to be twenty-two per cent thinner than specification and was to be subjected to pressure of from 135 to 140 pounds. It is a rule of the manufacturer when rebuilding a meter to replace the frost *279bottom “because of its age and because . . . presumably a certain amount of corrosion and weakening has taken place.” “ [Tjheir decision whether to put on a new one would depend entirely on the appearance of the bottom but ... if they were rebuilding a meter they would definitely put on a new bottom.”
It was undisputed that the defendant kept no records of meters and parts and what was done to them, a practice that, according to the testimony, was regarded as of high importance both by the manufacturer of the meter and by the defendant’s expert on municipal water systems.
There was testimony tending to show that the practice of the defendant in rebuilding meters, assuming an unused bottom was installed, was in accordance with good practice for municipal water departments, except in respect of the absence of records, also that it is not customary for municipalities to take pressure tests with gauges before installing water meters, and that the steps taken by the town to install the meter were usual and proper. The development engineer testified that it was perfectly proper to use meters made before the change in the specification for thickness of frost bottoms. The thickness of the frost bottom was determined by the plaintiff’s witness with a micrometer and it is not customary for municipal departments to have a micrometer. Calipers also could be used for such measuring.
There was also testimony that the meter was installed to replace a smaller meter when the plaintiff installed air conditioning. The auditor found that the defendant knew the purpose of the installation.
The plaintiff contends that it could have been inferred that the great deficiency in thickness of the frost bottom was due to wearing away from long use and abuse elsewhere, suggesting that this is a more reasonable inference than that the manufacturer had supplied a bottom so far below its specifications.
The inner surface of the disc is somewhat pitted but not deeply so and has a surface coating of rust on a part of it. An expert for the defendant called attention to “the origi*280nal black finish” on about one third of the surface, but testified also that it is “good practice to paint a used bottom with anti-corrosive paint before . . . [re-use].” There was no testimony that the appearance of the disc indicated erosion of metal. The plaintiff’s expert “didn’t find corrosion as a cause.” He testified that there was no definite evidence “that the rust formation along the sheared surface where the failure occurred developed previous to or following the failure.” The defendant’s experts testified, in substance, that the bottom looked like a new bottom, a part in use for only a relatively short time, and that there was no significant corrosion or rust, and that nothing in the appearance of the part would indicate that it should not have been used.
An employee of the defendant testified that in rebuilding this meter he used a new bottom “from their stock on hand” in accordance with practice. This contrasts with the auditor’s finding that although another bottom was put on when the meter was rebuilt the employees of the defendant did not know whether it was new or used.
The plaintiff reminds us that the judge’s findings “must stand unless they are unwarranted by any reasonable view of the evidence together with all rational inferences that may be drawn therefrom. His general finding is conclusive if there is any evidence to support it.” Kellogg v. Suher, 329 Mass. 544, 546. And, of course, the judge could disbelieve any part of the testimony.
Under these rules, we think that the judge, on some view of the evidence and within the limits of reasonableness, could have concluded that the defendant was negligent. True, the appearance of the frost bottom and most of the testimony about it tended strongly to show that the part had been in use over only a short period and to corroborate the testimony that a new frost bottom was used. It did not, however, require the judge so to find. This testimony did not make entirely insignificant either the extraordinary deficiency in thickness not accounted for by probable tolerance in manufacture or the auditor’s finding that the em*281ployees did not know whether the part was a new or used replacement. A conclusion that the part was made prior to 1944 could have some tendency to suggest the likelihood of some use in the intervening sixteen year period. The testimony of an experienced expert that he had never run into a case where a new frost bottom gave out in six weeks from graphitization also supported this conclusion. It was within reason for the judge to have concluded that weakness from graphitization existing at the time of installation and the relative thinness of the part were causes of the break. So far as appears, electrical conditions at the plant were equally operative on the meter that had been in place prior to July 14,1960.
We rule that the judge could have concluded that the defendant should have known the age of the frost bottom and whether it had had prior use and, in accordance with its own practice, and good practice, should have rejected the part if it had had any prior use. The judge could also have found that the defendant, not knowing the age of the part and whether it had had prior use should have rejected it, either out of hand, or in any event after inspecting and measuring it and finding that its critical thickness was sixty-two per cent under current specification. A test under the foreseeable pressure at the plaintiff’s plant could also have been found a requirement of due care, but it is speculative whether such a test would have shown the latent weakness of the part. Such cautionary steps may be required before using from stock a frost bottom of uncertain history even though, as the testimony suggested, good practice permits the continued use of old, long installed meters.
Great Atl. & Pac. Tea Co. v. Kennebec Water Dist. 140 Maine, 166, relied on by the defendant, also involved the failure of a frost bottom from a cause other than freezing. There, however, the meter had been rebuilt by the manufacturer. The court, finding that no amount of inspection by the defendant would have shown the defect, held that the doctrine of res ipsa loquitur was inapplicable, and ruled that the plaintiff had failed to make out a case.
*282We conclude that it was not error to deny the requests to rule that the defendant could not be held liable.
2. The possibility of a sustainable finding for the plaintiff does not, however, dispose of this case, for the judge’s ruling on the motion to strike the conclusion of negligence from the auditor’s report establishes that the judge may have acted on evidence and inferences inadequate to show negligence.
The auditor’s conclusion of negligence was not warranted on the subsidiary facts found by him. The auditor’s report is scanty. It described the frost bottom as so constructed “that a thin circular line was built in it which would give way or break when frozen.” The break “was not a clean break” and there was evidence of rust and corrosion. Tests “showed a thickness of twenty-two per cent less than the thickness of a new ‘frost bottom’ of the same type.” But there was no finding, other than of the break itself, to suggest that these factors were related to the break or, if so, that they indicated anything about the bottom when installed to tell the defendant that it should not have used the part. Other findings do not show negligence related to the break.3 The plaintiff argues that the auditor’s finding of *283negligence was inconsequential inasmuch as the judge, in reserving action on the motion, indicated an intention to decide the case in the light of all the substantive evidence, and we may assume that he would rest his conclusion on the complete view of the case unaffected by the auditor’s conclusion on a partial view of it. Perhaps the judge did this but we have no way of knowing. The denial of the motion is in effect a ruling that the subsidiary facts of the report warrant a finding of negligence. We may not assume that the judge did not apply to the evidence before him the view of the law manifested by that ruling. Doing so, he could have found negligence although rejecting the particular evidence, and not maldng the inferences that, as we have held, would have been necessary for such a finding. This possibility cannot be disregarded where, as here, some of the inferences necessary to support the finding, although within the limits of reason as we have held, are not strongly based and there is strong support for other, contrary inferences. If the intention was to decide the case on a basis apart from the only ruling made, specific findings or rulings to manifest this were reasonably required. Where an auditor’s report is in evidence before a jury, the giving of instructions is an appropriate method of eliminating erroneous statements of law in the report. Ferrairs v. Hewes, 301 Mass. 116,121. Kurland v. Massachusetts Amusement Corp. 307 Mass. 131, 142. Here, by contrast, the judge showed his affirmance of the erroneous conclusion of the auditor that the subsidiary facts found by the auditor warranted a finding of negligence. Having let stand the finding of negligence he may have rested his own finding of negligence on that ultimate finding. Cook v. Farm Serv. Stores, Inc. 301 Mass. 564, 566-567.
3. It is unnecessary to deal with the other contentions of error.
Exceptions sustained.
The diameter of the upper part of the broken frost bottom measures about four and three-quarters inches. The diameter of the base ring easting into which it fitted is about five inches. The record does not disclose what happened to whatever filled the space between disc and casting, but it can perhaps *276be inferred that the space was filled by a ring or ridge of metal, a part of the frost bottom, that disintegrated and washed away when the frost bottom broke. The plaintiff's expert, assuming that the bottom had had many years of service, indicated that there may have been rust accumulations (inferably at the line of the break although this was not' expressly stated) that had been dislodged by the rapid flow of water at the time of the break. The bottom was described as having a groove about its upper perimeter. There is some indication of this on the part exhibited' to us.
‘ ‘ [A] difference of 22 per cent less thickness in ■ the frost bottom that broke compared to another frost bottom would not be significant.”
The auditor found that water was furnished by the defendant to the plaintiff’s plant at a pressure of from 135 to 150 pounds per square inch and that the “meter was installed to withstand this pressure.” The meter was installed in late July or August, 1960, and was thirty to thirty-five years old, and had been used where the pressure was considerably less than at the plaintiff’s plant. This finding was based on mistaken testimony, corrected at the trial. The report stated that witnesses testified that another frost bottom was put on when the meter was rebuilt and that it was tested under the pressure at the Athol Water Works which was some 125 pounds per square inch. The auditor further found that the defendant kept no adequate records of repairs to and replacements in meters, and no record of the repairs on this meter. Also that the defendant's employees did not know whether the frost bottom was new or a used replacement. The night of September 2-3, 1960, when the meter broke was clear and warm. There was no evidence of freezing. The plaintiff had installed reducing valves within the building to keep down fluctuating water pressure or hammering. The defendant had not installed such valves outside the building although ‘ ‘ this is not uncommon in the municipality.” However, “it is not customary with towns to install reduction valves in their pressure systems.” After the break a hydraulic engineer conducted tests that showed variations in pressure from five to fifteen pounds and hammering as often as fifty times an hour. The fluctuations occurred from pressure outside the building.