OPINION
DANIEL F. FOLEY,
Judge
On April 3, 1997, appellant St. Paul Fire and Marine Insurance Company, as subro-gee of the Minnesota Indian Primary Residential Treatment Center (treatment center), filed a complaint alleging respondent Honeywell (1) breached its contract for preventive service, maintenance, and repair of the treatment center’s equipment and furnaces, and (2) negligently failed to properly service, maintain, and repair a furnace in the treatment center administration building. Appellant alleged respondent’s breach and negligence caused a furnace fire that resulted in $290,000 in damage to the treatment center. Respondent’s motion for summary judgment was denied and the case went to trial.
By special verdict form, the jury found the fire at the treatment center was directly caused by respondent’s (1) breach of its contract to service, maintain, and inspect the furnace, and (2) negligent failure to service, maintain, inspect, or repair the furnace. While the jury found the treatment center 20 percent contributorily negligent for the damages caused by the fire, the court initially ordered judgment for appellant and struck the jury’s 20 percent contributory negligence finding against the treatment center after considering appellant’s posttrial motions. Thereafter, respondent moved for JNOV or, alternatively, for a new trial.
Despite the jury’s special verdict in favor of appellant, the district court rejected appellant’s theory that respondent caused the October 14, 1995 fire by failing to properly tighten a pump flare fitting in November 1994 and perform preventive maintenance service in the fall of 1995.
With regard to the loose flare fitting, the district court found there was no competent evidence to support appellant’s assertions and that any evidence that respondent altered or loosened the flare fitting before the fire was “speculative and based improperly on the occurrence of the accident.” In addition, the district court concluded that appellant’s theory was .contrary to the uncontroverted evidence that the furnace did not leak or malfunction between November 1994 and 1995. The district court also rejected appellant’s claim that respondent failed to provide preventive maintenance service because the contract had no specified date for such service, and the claim is speculative and improperly based on the occurrence of the fire itself. Concluding there was no competent evidence to sustain the verdict, the district court granted JNOV.
Appellant challenges the district court order granting respondent’s motion for JNOV or alternative motion for a new trial. Because there is sufficient evidence to sustain the jury verdict, we reverse and reinstate the jury verdict.
FACTS
In 1980, the treatment center contracted with respondent to provide “preventive maintenance,” replace components that are “worn, failed, or doubtful,” and provide emergency service. The contract also required respondent to make a minimum of “eight visits per year,” perform periodic maintenance, and “do seasonal start-ups in the spring and fall.” Following the addition of the administration office building in 1988, respondent agreed to service the administration building furnace in addition to the treatment center’s other four furnaces.
Jeff Johnson, the treatment center’s business manager, testified that only respondent serviced the furnaces. Jerry Lundberg is the service representative assigned to service and maintain the treatment center’s furnaces. As reflected by Lundberg’s service reports, he typically performed preventive maintenance on the furnaces in September and October, but the contract did not specify a particular date for such service. The service reports indicate that annual fall preventative maintenance was performed each year except 1995.
On November 3, 1994, Lundberg was called to service the administration building furnace because it had a “funny odor” and “didn’t seem to be burning properly.” Lundberg’s service report indicates “blow back on Adminstration furnace” due to lottery tickets blocking the chimney elbow. Linda Bubacz, a housekeeper for the treatment center, called Lundberg on November 11, 1994, after detecting no heat in thfe administration building, smelling oil in the furnace room, and observing a puddle of oil on the floor. Lundberg’s November 11, 1994,' service report indicates there was “no heat in office area,” the “pump couplings [were] stripped,” and he “replaced [the] couplings.” While the parties dispute whether “office' area” meant the administration building or offices in the treatment center, Johnson testified that Lundberg meant the “office area in the administration building,” and Bubacz testified that she took Lundberg to the administration building furnace room.
In order to fix a stripped pump coupling, Lundberg testified that he had to remove the pump. Lundberg explained removing the pump required him to loosen the upper flare fitting on the pump, loosen the fitting on the supply line (the inlet supply of pressure to the pump), and loosen set screws fastening the pump to the burner housing. After loosening those items, Lundberg testified that the pump slides right out and that nothing has to be done to the fitting on the output tube attached to the pump. He explained that after replacing the stripped coupling he restarted the furnace and observed it in operation to determine if it was working properly.
While Lundberg returned to the treatment center on at least three occasions during the 1994-95 heating season for rou
tine inspection of the furnaces, he noted nothing out of the ordinary. Johnson, who entered the furnace room twice daily to turn lights on and off, did not report any malfunctioning or problems prior to the fire. Similarly, Bubacz did not detect the smell of fuel oil after Lundberg’s November 11, 1994 service call, despite entering the furnace room weekly for cleaning supplies.
On October 14, 1995, Karen Lindholm noticed black smoke in the administration building while retrieving a vacuum cleaner. Lindholm reported the fire to fellow employees Keith Lind and Curt Rainey who sounded the fire alarm. Rainey and Lind went to the administration building with fire extinguishers and discovered the fire after opening the furnace room door. Rai-ney testified that he simply saw a big flame until he unloaded the fire extinguisher on the flame and saw the furnace was burning. Other than the furnace, Rainey saw nothing else burning in the furnace room. While the record does not indicate whether Rainey closed the furnace room door before leaving, his trial testimony shows he hurried to leave the furnace room.
Fire departments from Perch Lake and Cloquet responded to the fire. Cloquet fireman, Brad Juntunen, testified at a May 22, 1998 deposition that he remembered the furnace room door being open, but also that he recalled having to pry open the interior and exterior doors to the furnace room. While Juntunen did not recall using foam to fight the fire, he explained the inside crew was using a line from the Perch Lake fire engine and only knew that Cloquet’s engine did not use foam. Instead, Juntunen explained that the firemen used traditional water and ventilation methods to extinguish the fire and testified that the fire was difficult to extinguish because it rekindled itself several times. The evidence shows that the Perch Lake Fire Department billed the treatment center for the use of foam and some firemen recall using foam, while others do not. Steve Olson, Perch Lake Volunteer Fire Department’s fire chief, reported that the fire originated in the furnace room based upon his review of the burn patterns and interviews with the firemen. Olson arrived after the fire had been extinguished. Therefore, his conclusion that fuel oil had ignited was based on the firemen’s statements regarding the difficulty of extinguishing the fire and the use of class b foam, which is typically used to extinguish combustible liquid fires. Because firemen reported having to shut off a valve controlling the flow of Duel oil, Olson also concluded the fuel line failed as a result of the fire and fuel oil continued to pump from an underground tank.
Mark Germain, a fire investigator from the State Fire Marshall’s Office, investigated and photographed the fire scene the day of the fire. Germain found no evidence of leaking oil lines to the furnace, oil on the floor or anything combustible burning at or near floor level between the furnace and the furnace room door. Based on his burn-pattern analysis and witness statements, Germain concluded the fire originated inside the furnace or right outside the furnace. Germain also testified there was no lower burn damage to substantiate a fuel oil leak except towards the front of the furnace and the burn patterns corroborate Rainey’s statement that the fire came from inside the furnace. Crawford Wiestling, a fire cause and origin investigator, visited the fire scene on October 18, 1995. During his initial investigation, Wiestling photographed the furnace room, furnace room door, and debris, and took smudge samples from various locations in the structure. On October 27, Wiestling and Sid Bhatt, a professional engineer specializing in oil furnace fires, examined the fire scene and Wiestling reported checking the fuel oil valve. Following their on-site investigation, Wiestling and Bhatt loaded the furnace on a trailer and transported it to a storage locker in Prior Lake.
In July 1996, potentially interested parties removed the furnace from the storage locker for inspection. The burner was removed during this inspection. At a second inspection in September 1996, the pump and transformer were tested and leaks were found at the shaft seal and pump gasket. Following a decision to conduct additional testing on the burner in Plymouth, Bhatt discovered a loose flare fitting. In June 1997, Bhatt and Honeywell representatives returned to the treatment center to further investigate the fire scene and other oil furnaces.
Following his investigations, Wiestling found no evidence of oil residue on the floor and a lack of lower burn damage to the furnace room door, interior doorframe, and hallway. Wiestling testified that the fire damage was limited to the top of the furnace room door and “high tracking” across the tops of adjacent doors. Wies-tling noted evidence of low burning only between the furnace and the door and full radiant heat impingement on the inside of the furnace room door. Additionally, Wiestling concluded there was no fuel oil ablaze on the floor in front of the furnace because the casing on the pump was not oxidized, the fuel housing “still retained its bright shiny color,” and the electrical conduit behind the wall had not been affected by high heat at floor level. Based on his investigations, observations, and the evidence, Wiestling concluded the fire started inside the furnace and traveled by “radiant energy” out the furnace’s inspection port.
On October 27, 1995, Bhatt visually inspected and photographed the furnace and furnace room and disassembled electrical components to inspect the circuits. While Bhatt did not observe fuel oil residue on the floor or much evidence of floor-level fire damage, he observed significant heat damage to electrical wire insulation at elevated locations. Bhatt also testified that the soot he found in the combustion chamber was abnormal for a “properly operating furnace.” During the joint inspections of the burner, Bhatt found a loose flare fitting while loosening the fitting connecting the oil tube to the pump and carbon deposits around the loose flare fitting. Based on his examination of the burner and observations of the furnace and furnace room, Bhatt opined the fire was caused by delayed ignition of the furnace from oil leaking around the loose fitting.
■ Respondent’ retained Richard “Smokey” Dyer, a fireman with 25 years experience investigating fires, to determine the origin of the fire, its probable ignition source, what material was originally ignited, and the mode of propagation. To render his conclusions, Dyer reviewed physical evidence collected from the- furnace room and reports prepared by the Cloquet and. Perch Lake Fire Departments, the State Fire Marshall’s office, and the independent fire investigator. Dyer also reviewed numerous photographs and the deposition testimony of treatment center employees, eyewitnesses, and experts. At trial, Dyer admitted he is neither a furnace expert nor qualified to testify as to how the oil leaked or was ignited.
Respondent also hired John Forss to determine the cause of the fire. However, Forss, a 30-year Honeywell employee, testified at trial that he does not consider himself an expert in determining the cause and origin of fires. Forss currently owns his own consulting firm and handles 100-150 claims annually for respondent. Forss attended the joint inspections of the furnace and burner in July 1996 and the disassembly, inspection, and testing of the burner in September 1996. To render his conclusions and opinion, Forss reviewed the fire reports, maintenance service reports, statements, depositions, and manuals from other furnace burner manufacturers. In addition, Forss visited the treatment center in June 1997 to look at the facility’s other oil'furnaces and pumps to help visualize the system prior to the fire.
ISSUES
I. Did the district court err in granting respondent’s motion for JNOV?
II. Did the district court err in further ordering that respondent’s alternative motion for new trial be granted in the event JNOV is reversed or vacated?
ANALYSIS
I.
A district court’s grant of JNOV raises a question of law, which is reviewed de novo.
Diesen v. Hessburg,
455 N.W.2d 446, 449 (Minn.1990). JNOV should not be granted unless the evidence, viewed in the light most favorable to the verdict, is “practically conclusive against the verdict and reasonable minds can reach only one conclusion.”
Id.
at 452 (citation omitted). However, a district court may grant JNOV and set aside a special verdict when the evidence does not sustain the verdict or the jury’s findings are contrary to the law.
Id.
In this case, each party presented a theory of causation and evidence in support of its position.
Causation Theories
Appellant’s furnace-fire expert, Sid Bhatt, opined Lundberg’s failure to tighten the flare fitting after servicing the furnace in November 1994 directly caused the fire. Because a loose flare fitting would cause oil to leak or spray while the furnace was running, Bhatt explained that oil would be sucked into the burner tube and coat the porcelain insulators and ends of the electrodes with a sticky buildup of dirt, lint, and particles. Bhatt testified that electric current would flow through the contaminated materials on the insulators, bypassing the high voltage output that should be going to the electrodes to produce a spark. He explained that electric current flowing through the contaminants could short out the furnace or cause arcing and sparking on the insulator surfaces. Until the contaminants are burned off and the spark returns to the nozzle, Bhatt explained that furnace ignition would be delayed, causing a larger than normal volume of oil vapor and air to collect. Concluding the resulting ignition could be a puff, bang, or worse, Bhatt opined that delayed furnace ignition caused the fire.
Alternatively, respondent argued the fire was caused by an oil leak, resulting from a sudden failure of the pump seal after treatment center personnel left the furnace room on October 13, 1995. Respondent contended oil vapors ignited in the normal operation of the furnace after oil had accumulated in the furnace and spilled onto the furnace room floor. In support of its “sudden failure” causation theory, respondent presented the testimony of Richard “Smokey” Dyer and John Forss.
Based on his examination of photographs two and a half years after the fire occurred, Dyer found no evidence of radiant heat impinging the furnace room door at the level of the furnace inspection port, excessive heat damage to the furnace’s interior, or combustible materials in the furnace room to be ignited by radiant heat. Instead, Dyer found low burn damage in the southwest corner of the furnace room and significant damage to the entire inside of the furnace room door.
Based on the low-level burn patterns in the southwest corner of the furnace room, the firemen’s use of foam and their difficulty in extinguishing the fire, and the lack of other combustible materials in the furnace room to support the fire, Dyer opined fuel oil vapors ignited in the furnace vestibule and heated the spilled oil on the concrete floor to its flash point, spreading the fire throughout the furnace room. While Germain and Wiestling did not find evidence of fuel oil on the furnace room floor, Dyer still concluded it was a fuel oil fire on the floor because the fire would have consumed most of the fuel oil and there was no absorbent material to retain the fuel oil vapors.
Dyer testified that the burn patterns were consistent with a two-gallon fuel oil fire and the proper concentration of fuel oil vapors will ignite around 110 degrees Fahrenheit. The evidence shows it was 30-40 degrees outside on October 14, 1995
and on-scene inspectors did not detect fuel oil on the floor. Moreover, Dyer did not test or inspect similar furnace models to determine if the front wall of the furnace or the concrete floor could reach 110 degrees and admitted he is neither a furnace expert nor qualified to testify as to how the oil leaked or was ignited.
Dyer rejected appellant’s theory that the fire communicated throughout the room by radiant heat energy. Because the aluminum insulation inside the furnace was not melted, despite its melting temperature of 1200. degrees, Dyer concluded the furnace did not reach 2000-2500 degrees Fahrenheit or give off radiant heat rays. Although Rainey and Lind said nothing about seeing a fire on the floor and reported seeing and fighting the fire inside the furnace, Dyer interpreted their deposition testimony to mean they discharged the extinguishers in the direction of the furnace. Other than considerable damage to the furnace, Dyer found no evidence consistent with an internal furnace fire and rejected all other possible fire causes.
Forss attended the joint inspection where Bhatt discovered the loose flare fitting. Forss opined Lundberg’s service in November 1994 did not cause the loose flare fitting to leak because the furnace went through the 1994-95 winter heating season without incident. Instead, he concluded the fitting became loose on the day of the fire because a loose fitting prior to the fire would have dripped fuel oil and nobody detected the smell of fuel oil. While Forss could not explain how the flare fitting became loose, he believed it might have loosened during a knock down of the room following the fire or during transportation for off-site inspection of the furnace and burner.
Because the loose flare fitting and fuel oil leak were not detected, Forss opined some external, unpredictable event, like sudden wear or failure of a component, caused the properly serviced oil furnace in the administration building to leak oil and start the fire. While Dyer testified that fuel-oil vapors must reach 110 degrees to ignite, Forss explained that the oil leaking from the pump into the furnace vestibule is hot enough to light the vapors. Therefore, Forss concluded a sudden increase in pressure in the oil line caused the oil pump seal to blow, resulting in a continuous oil leak into the lower compartment of the burner and spilling onto the concrete floor.
Forss disagreed with Bhatt’s delayed ignition theory because he did not believe that (1) the flare fitting at the end of the output tube would have to be loosened to replace a stripped coupling; (2) oil would spray out of the loose flare fitting; and (3) oil coating the electrodes would result in delayed ignition. Because a pump leak, unlike a loose fitting leak, would cause the continuous flow of oil even if the furnace shut down, Forss rejected appellant’s causation theory. Even if the oil was sucked into the burner’s combustion chamber, Forss testified he would still conclude a pump, leak caused-the fire because arcing and sparking would not fool the furnace’s flame sensor and burner control, which would have shut down the furnace when ignition did not occur within the set time period.
Review of the Verdict
By special verdict form, the jury found respondent directly caused the fire by negligently failing to service, maintain, inspect or repair the treatment center furnace, and breaching its service and maintenance contract.. In granting JNOV, the district court concluded appellant’s causation theory was speculative and not supported by competent evidence. In considering a motion for JNOV, a district court “must view all the evidence in the light most favorable to the verdict” and “may not weigh evidence or judge witness credibility.”
B & H Inv. Co. v. Union Stockyards Co. of Fargo,
513 N.W.2d 264, 267 (Minn.App.1994),
review denied
(Minn. May 17, 1994). If a jury verdict has any reasonable evidentiary support, both the trial court and appellate court must accept it as final
Brubaker v. Hi-Banks Resort
Corp.,
415 N.W.2d 680, 683 (Minn.App.1987),
review denied
(Minn. Jan. 28, 1988). De novo review requires this court to examine the record to determine whether the evidence can support the special jury verdict, and is consistent with the law.
Diesen,
455 N.W.2d at 452.
Appellant’s causation theory is supported by evidence showing that (1) Lund-berg replaced a pump coupling on the administration building furnace on November 11, 1994, (2) Lundberg had to remove some flare fittings to replace the pump coupling; (3) a loose flare fitting leaks oil; (4) a loose flare fitting can be discovered by proper inspection; and (5) Bhatt discovered carbon deposits on the sides of a loose flare fitting. There is expert testimony that a properly tightened flare fitting will not loosen by itself even under the extreme conditions of a fire and no other flare fittings were found loosened from the fire or transportation. Respondent’s service contract does not specify a date for fall maintenance of the furnaces, but the evidence shows Lundberg provided such service in September and October each year except 1995, the year of the fire. While Lundberg testified about his maintenance service on numerous other occasions, he was unable to recall his November 11, 1994 service of the administration building furnace.
As additional support for its theory that the fire started inside the furnace, appellant presented evidence and photographs illustrating more extensive burn damage to the upper portion of the walls, door, and doorframe of the furnace room. After investigating the fire scene, Germain and Wiestling each concluded it was an internal furnace fire rather than a floor fire because the burn patterns showed less lower burn damage. In addition, both parties agree no fuel oil residue was detected on the floor. But while appellant contends the lack of detection is evidence of respondent’s negligent maintenance and breach of the service contract, respondent argues the fuel oil was burned off by the fire or that there was no loose flare fitting or oil leak to be detected. This conflict in the evidence was for the jury to resolve.
Moreover, appellant argues the lack of damage on the exterior side of the furnace room door is consistent with an internal furnace fire because a fire on the floor would have damaged the lower portion of both sides of an open furnace room door. While the parties dispute whether the furnace room door was open during the fire, the evidence is inconclusive. Appellant contends it is reasonable to infer Rainey left the furnace room door open because he testified he was in a hurry to get away from the fire. Cloquet fireman Juntunen initially testified he believed the door to the furnace
room
was open when they arrived, but later recalled firemen having to pry open both doors to the furnace room.
Appellant also contends it was an internal furnace fire because Rainey and Lind said they discharged fire extinguishers on the furnace and saw flames coming up from the furnace’s interior. In contrast, respondent argues it was a combustible fuel-oil fire on the floor because the fire was fought with foam. While the parties dispute the use of foam, the evidence shows that the Perch Lake Fire Department billed the treatment center for the use of foam, the Cloquet fire engine parked behind the building did not use foam, and some fireman recalled using foam, while others did not.
Each party presented its causation theory. Respondent only objected to Bhatt’s testimony regarding whether its negligent maintenance and failure to inspect the administration building furnace directly caused the fire. While there is conflicting expert opinion testimony and evidence regarding causation, the record amply supports the jury’s special verdict findings that Honeywell caused the fire by negligently failing to service and maintain the furnace and breaching its service and maintenance contract. The central issue as to causation — the start of the fire — was
a jury question. Because we cannot conclude the evidence is practically conclusive against the verdict such that reasonable minds can reach only one conclusion, JNOV was improper.
Id.
In addition, the jury’s verdict is consistent with the law. It is within the province of the jury to weigh conflicting testimony and determine witness credibility.
Tsudek v. Target Stores, Inc.,
414 N.W.2d 466, 469 (Minn.App.1987),
review denied
(Minn. Dec. 13, 1987). Causation is a question of fact for the jury and the jury’s finding will not be set aside unless it is manifestly and palpably contrary to all the evidence viewed in the light most favorable to the verdict.
Savage v. K-Mart Corp.,
393 N.W.2d 25, 27 (Minn.App.1986). In concluding appellant failed to establish causation, the district court impermissibly weighed evidence and judged witness credibility.
See B & H Inv. Co.,
513 N.W.2d at 267 (requiring district court to view evidence in light most favorable to verdict without weighing evidence or judging witness credibility). A verdict is not rendered contrary to law simply because the district court believed the evidence supported an alternative result.
See Covey v. Detroit Lakes Printing Co., a
Div,
of Forum Pub. Co.,
490 N.W.2d 138, 142 (Minn. App.1992) (explaining jury’s answers will not be set aside unless perverse and palpably contrary to the evidence);
Tsudek,
414 N.W.2d at 469 (refusing to disturb special verdict if it is reconcilable on any theory). Because there is competent evidence to sustain the verdict and causation presents a question of fact for the jury, not a question of law for the court, the district court erred in granting JNOV.
II.
The trial court further ordered that in the event JNOV was reversed or vacated, respondent’s motion for new trial was granted because the “jury’s verdict is not justified by the evidence and is contrary to law.” A district court’s discretion to grant a new trial will not be disturbed absent a clear abuse of that discretion.
Halla Nursery, Inc. v. Baumann-Furrie & Co.,
454 N.W.2d 905, 910 (Minn.1990). However, a district court should not grant a new trial
unless the verdict is so contrary to the preponderance of the evidence as to imply that the jury failed to consider all the evidence or acted under some mistake or from some improper motive,bias, feeling or caprice, instead of honestly and dispassionately exercising its judgment. • •'
Lamb v. Jordan,
333 N.W.2d 852, 855-56 (Minn.1983) (citation omitted);
see also Eliason v. Textron, Inc.,
400 N.W.2d 805, 807 (Minn.App.1987) (explaining new trial motions should be granted cautiously and reluctantly). Here, the evidence justifies the verdict and there is no indication the jury failed to consider all the evidence or acted under some mistake, improper motive, bias, or caprice. Therefore, we must also reverse 'the district court’s order granting respondent’s alternative motion for new trial.
In its reply brief, appellant stresses that “the case essentially turned upon the credibility of the witnesses.” We agree. Appellant seeks reinstatement of the judgment entered upon the special verdict, which we grant by this decision. The effect of this decision is to reinstate the 20 percent finding of fault entered against the treatment center.
DECISION.
The district court improperly granted JNOV. because the jury’s special verdict findings that respondent’s negligence and breach of contract .caused the fire were supported by the record and consistent with the law. The district court improperly granted respondent’s alternative motion for new trial because there was sufficient
evidence to sustain the verdict and no indication of juror misconduct was shown. Thus, we reverse and remand with instructions to reinstate the jury verdict and to otherwise proceed in accordance with this decision.
Reversed and remanded.