On 12 January 1974 a fire completely destroyed eight units of a motel located on the outskirts of Minot. The district court concluded the fire was caused by the malfunction or faulty design of an electric wall heater manufactured and supplied by McGraw-Edison Company and installed by Plains Electric, Inc. Judgment was entered accordingly, from which both McGraw-Edi-son and Plains Electric appealed.
In 1973, the plaintiffs, Lawrence and Pat Schmidt, were the owners of a motel business called Pat’s Motel. The units of that motel were heated by propane gas. In response to a rumored future shortage of propane, Pat Schmidt sought the installation of electric baseboard heaters into each of the units to serve as supplemental or alternative heat sources. She contacted Plains Electric which agreed to sell and install heaters manufactured by McGraw-Edison. The project of installing the heaters was completed on 26 December 1973. The heaters were thermostatically controlled and two such heaters were placed in each unit. The president of Plains Electric testified he instructed his installation team to place the heaters one inch off the floor and to center them beneath a window. The two men who installed the heaters testified they complied with those installation instructions as much as possible.
Plaintiffs alleged Unit 16 was the unit in which the fire started. At the time of the fire Unit 16 was occupied by Allen L. Binder, a Minot school teacher. Testimony showed Binder had lived in the unit for a period of time prior to the fire. Binder’s practice was to directly leave school on Friday afternoon for his parents’ home in another community, and return to the motel on Sunday evening.
Binder testified that after the electric heaters were installed, he closed the registers for the propane furnace and relied on the baseboard heaters as the sole source of heat. Accordingly to Binder, on the Friday before the fire he awoke at approximately 2:30 a. m. because of excessive heat in the room. He said he turned down the thermostat and opened the doors to the unit until it cooled off and then he went back to sleep. The next morning the room was comfortably warm. Binder set the thermostat and *797left for school and in keeping with his usual practice did not return to the room that Friday afternoon.
On another occasion, approximately a week before the fire, the two motel maids found the same Unit 16 excessively warm to the point they could hardly touch the furniture. They notified Mr. Schmidt, who turned down the thermostat and the room cooled off.
Another resident of the motel, Agnes Le-Sarge, testified that on the morning of the fire she was sleeping in Unit 15. She awoke and smelled something like scorched paper coming from the gas heater. The smell grew stronger. She decided to report it and while she was on her way a lady from Unit 17 started yelling “Fire.”
Patsy Schmidt, the daughter of the owners of the motel, testified that when she heard the lady from unit 17 yelling she opened the door to unit 16 and saw white smoke. When she returned to Unit 16 a second time with a fire extinguisher, there was more smoke but still no flames. She then shut off the electric power to those units. Shortly after that the Minot rural fire department arrived but was unable to extinguish the fire before it had consumed Units 14 through 21.
The Schmidts commenced an action against Plains Electric. Binder and two other occupants also initiated an action against Plains Electric seeking damages for the destruction of their personal property. McGraw-Edison became a third-party defendant to both actions after Plains Electric filed a third-party complaint. The cases were consolidated and tried to the district court sitting without a jury. The trial court found that “As a result of the malfunction and faulty design of the heater in Unit # 16, excessive heat was generated and built up, causing ignition to occur in Unit # 16.” The court concluded that McGraw-Edison was liable for all the damages suffered by the plaintiffs. McGraw-Edison filed a consolidated motion for a new trial, amendment of findings of fact, and additions to the findings of fact. The motion was denied and judgment was filed. McGraw-Edison appealed from the final judgment and the order denying the motion for a new trial and amendment of findings. Plains Electric appealed from parts of the final judgment.
I
Although other collateral inferences were raised during the course of trial, the plaintiffs relied primarily on one theory in seeking to establish liability on the part of Plains Electric and McGraw-Edison. The basis of their theory was that the heaters had been installed behind some acrylic-backed drapes. The plaintiffs contended, in effect, that because of the malfunction of safety switches on the heaters, or because of the way the heaters were designed in the placement of the safety switches, excessive heat was allowed to build up between the heaters and the drapery material until the ignition point of the drapery material was reached. In support of that theory, the plaintiffs offered the testimony of an expert witness, Sharad Bhatt. Bhatt testified he conducted tests on both the drapery material and the heaters. In brief, the main significance of those tests was that in a normal operating condition, without blockages, the heater was capable of producing temperatures of between 301° and 310° F., and with various parts of the heater blocked, such as the air inlet or outlet, the heater produced temperature ranges of up to 411° and 447° F. He also testified the drapery material, when folded to facilitate spontaneous combustion, had an ignition range, based upon various exposure times, of between 409° and 459° F.
McGraw-Edison argued on appeal that if the plaintiffs’ theory is accepted several assumptions must be made, such as the position of the heater in relation to the drapes, the effect this position would have on restricting air flow, and the function or malfunction of the thermostat. McGraw-Edison contended the weight of evidence on several of these assumptions is directly contrary to and does not support the plaintiffs’ theory.
*798Initially we disagree with McGraw-Edison’s contention that all the factors it set forth are absolutely necessary to support the plaintiffs’ theory of how the fire started. As to those factors which were necessary to reach the plaintiff’s theory, evidence was offered to support each one. It is important to note very little of the expert testimony that was offered, as well as the testimony concerning the setting of the room, was uncontradicted. It would, however, serve little purpose to recite the testimony supporting the plaintiffs’ position. Suffice it to say, sufficient evidence was offered and received from which inferences in support of the plaintiffs’ theory could be drawn. Inferences drawn from the evidence may constitute the basis for the findings of fact made by the trial court. Slope County, Board of County Commissioners v. Consolidation Coal Co., 277 N.W.2d 124 (N.D.1979). In reviewing findings of fact this court is governed by the clearly erroneous rule of Rule 52(a), North Dakota Rules of Civil Procedure. Under this rule we will set aside a finding of fact only when, although there is some evidence to support it, we, after reviewing the entire evidence, are left with a definite and firm conviction a mistake has been made. In re Estate of Elmer, 210 N.W.2d 815 (N.D. 1973). In applying this rule, we set findings of fact aside only if they are found to be clearly erroneous based upon all the evidence, and not merely because we may have reached a different result had we tried the case. Anderson v. Mooney, 279 N.W.2d 423 (N.D.1979). What McGraw-Edison, Plains Electric, or even this appellate court might view as the greater weight’of the evidence concerning the operation of the safety switches, the relative position of the drapes to the heater, the effect of this position on restricting air flow, and the function or malfunction of the thermostat has little relevance when the record contains sufficient admissible evidence to support the trial court’s findings as there was in this case.
II
McGraw-Edison argued the trial court failed to make findings of fact on the critical points set forth above. It contended, in effect, that without such findings, this court is not afforded a clear understanding of the trial court’s decision. McGraw-Edison further argued the trial court’s failure to include these determinations in its findings of fact indicates the facts were never ascertained by the trial court. We agree the findings of fact could have been more definitive, but this still does not make them erroneous. McGraw-Edison also criticized the practice employed here of having the prevailing party prepare findings of fact and conclusions of law,1 but neither does this criticism, when applied to the findings of fact in this case, lead us to a definite and firm conviction that a mistake has been made so as to set aside those findings.
The trial court made the following pertinent findings of fact:
“The fire on January 12, 1974, at the motel property owned by the plaintiffs was the result of the malfunction and/or faulty design of the electric wall heater installed in Unit # 16 by Plains Electric, Inc., and manufactured and supplied by McGraw-Edison Company .
“As a result of the malfunction and faulty design of the heater in Unit # 16, excessive heat was generated and built up, causing ignition to occur in Unit # 16.”
McGraw-Edison would have us set aside the findings because factual questions nec*799essary to reach them were not recited. Although such a recitation would have been appropriate and helpful, the failure to make the recitation does not constitute error of the dimension so as to warrant setting aside the findings of fact and remanding this case.
Because the plaintiffs offered basically one theory in way of explaining the origin of the fire, we must assume that when the trial court accepted that theory and made a factual finding to the same effect, it also accepted and found credible the evidence in support of that theory and the inferences drawn therefrom. We have already stated the plaintiffs’ expert testified regarding the ignition temperatures of the drapery material and that the heater was capable of producing those temperatures. There was also testimony from which it could be inferred that the heater was off-centered and that the drapes could have obstructed the free flow of air over the heater. In addition, testimony was offered from which it could be inferred that the fire started in the vicinity of the heater, such as the spread of the flames, excessive heat which appeared to originate from the heater and which was detected in the room on two prior occasions, excessive damage to what appeared to be a wall heater found in the vicinity of Unit # 16, and the condition in which electrical wiring to that heater was found. This is a simplified summary of only some of the evidence that was offered, most of which was highly technical. We conclude, however, such evidence was sufficient to support a finding by the trial court that the fire resulted from the malfunction or faulty design of the wall heater even though some technical and some non-technical testimony was offered to refute this. Even if we were to agree with the appellees as to the weight of the evidence on certain points, questions of fact decided by the trial court upon conflicting evidence are not subject to re-examination by this court. Dickinson Education Association v. Dickinson Public School District No. 1, 252 N.W.2d 205 (N.D.1977). Neither is it our function to provide expertise on technical, non-legal -subjects, nor to appraise the credibility of expert witnesses. Foremost Insurance Co. v. Rollohome Corporation, 221 N.W.2d 722, 728 (N.D.1974).
Ill
McGraw-Edison also argued that if the trial court found, which it apparently did, that the heaters were installed under the drapes, then the plaintiffs’ allegation of negligent installation by Plains Electric would have been established and no third-party indemnification could be allowed against McGraw-Edison. McGraw-Edison also argued the trial court could not find the heater malfunctioned to the point of allowing an excessive heat buildup, unless it also found the thermostat was negligently installed or non-operational, in which event either Plains Electric or the thermostat manufacturer would be liable for the fire.
We do not find evidence which required the trial court to conclude the thermostat was non-operational or negligently installed if it found excessive heat buildup in or near the heater. The trial court could have determined that because of the extremely cold outside temperatures, it was possible for the drapery material to be subject to excessive heat without the room reaching a temperature which would cause the thermostat to shut off the heater.
We also do not accept the argument that if the heaters were installed behind the drapes, Plains Electric must be found negligent in the installation of the heaters and no third-party indemnification could be allowed against McGraw-Edison. The instruction booklet McGraw-Edison furnished with the heater in question stated:
“IMPORTANT! To obtain the maximum efficiency and economy of operation from any baseboard heating system, it is important that there be free circulation of air through the units. Do not block the air inlet (bottom! or air outlet (topi with draperies, rugs, furniture or other materials that will prevent or restrict the free flow of air.” [Underscored in original.]
*800The above instruction pertains only to providing for optimum conditions for the efficient operation of the heater. It does not warn of a potential fire hazard. The president of Plains Electric, as well as the two men who installed the heaters, testified they were never informed that installation of the heaters behind drapes could create a fire hazard. To establish negligence on the part of Plains Electric, as a supplier of the product, it would have been necessary to show knowledge or notice of the defect in, or the dangerous situation created by the product. Continental Casualty Co. v. Belknap Hardware & Manufacturing Co., 281 S.W.2d 914 (Ky.1955); 72 Supp. C.J.S. Products Liability § 16. There was no showing of such knowledge of notice in this case. Under the facts presented we will not hold the supplier liable for negligent installation, where the instruction provided by the manufacturer pertained to the operating efficiency of the heater but did not provide knowledge or warning of a potential fire hazard.
IV
Plains Electric raised the argument that the tests conducted by Bhatt, and apparently relied upon by the trial court in reaching its findings, were not substantially similar to the conditions as they allegedly existed at the time of the fire. Plains argued that in North Dakota an experiment or test is admissible as evidence only where the conditions surrounding the test are substantially the same as those prevailing at the time of the occurrence to which the test relates. Plains stated that Bhatt conducted his tests with a bedspread covering the center of the heater while there was no evidence of the heater in Unit 16 being covered by a bed-spread. In addition, Bhatt did not conduct any tests placing the drapes and heater in the configurations the Schmidts stated existed at the time of the fire.
This court has stated that the foundation testimony offered in support of the admission of the results of tests must establish that the conditions surrounding the tests were substantially the same as those prevailing at the time of the occurrence to which they relate. VanOrnum v. Otter Tail Power Co., 210 N.W.2d 188, 197 (N.D.1973). The admission of evidence of such experiments and tests is primarily within the sound discretion of the trial court, the exercise of which will not be disturbed on appeal unless an abuse of discretion exists.
“However, in exercising its discretion, the trial court should always be aware of the principal object to be achieved by the admission into evidence of results of tests, and that is the ascertainment of truth with reference to the existence or nonexistence of facts in controversy. 32 C.J.S. Evidence § 587, pp. 715-717.” Van Ornum v. Otter Tail Power Co., supra at 197.
The foundation testimony initially offered by the plaintiffs in support of the admission of Bhatt’s test results prompted the trial court to say that although it would receive the evidence, it was bothered by the dissimilarity of the tests and the conditions prevailing at the time of the fire.
This court has stated that:
“We believe that a trial judge, in a nonjury case, should ordinarily admit all evidence which is not clearly inadmissible. A judge who is competent to rule upon the admissibility of evidence can distinguish in his own mind, when deliberating his ultimate decision, between evidence which is admissible and evidence which is not admissible.” Tallackson Potato Co., Inc. v. MTK Potato Co., 278 N.W.2d 417, 423 (N.D.1979); Schuh v. Allery, 210 N.W.2d 96, 99-100 (N.D.1973).
The above rule is particularly applicable in this case. As Bhatt’s testimony was developed, further foundation was established as to its relevancy to the issues of the case. Our rule requiring similarity of circumstances does not dictate a duplication of circumstances. Similarity of circumstances in Bhatt’s tests consisted of the types of heaters and drapes used in the test and those alleged in Unit 16 at the time of the fire. This similarity was sufficient to allow the trial court, within its discretion, to admit the results and then rely on those *801parts of the tests relevant to the issues in reaching its determination. We conclude sufficient similarity of circumstances existed between the tests and the alleged conditions before the fire, that the admission of the results of those tests was not an abuse of discretion.
V
Another collateral issue raised by McGraw-Edison concerns the following statement in the trial court’s memorandum decision:
“The fire at the motel was thorough and devastating, so there was little remaining that would help solve the problem. However, one expert [Bhatt] did examine what he thought was the remains of the electric wall heater from Unit # 6. He testified that his examination showed some of the fins on the heater had been subjected to extreme heat which would would have to have come from the heater itself.”
Bhatt testified he found the remains of what he thought was the wall heater from Unit 16 and that his examination showed the fins of that heater had been subjected to extreme heat. He also stated, however, that the 1100° temperature that would have been necessary to damage the fins could not have been generated by the heater itself. Rather, he testified the damage to the fins led him to believe the heater had been subjected to greater heat from independent fuel sources than the heaters in the other units which did not sustain such damages. In his opinion, the excessive damage to the fins of the heater found in Unit 16 indicated the fire originated near this heater.
McGraw-Edison argued the trial court’s misstatement of the evidence demonstrated a basic misunderstanding of that evidence which affected its underlying findings of fact. We do not agree with this argument.
McGraw-Edison properly pointed out the discrepancy to the trial court in its brief in support of its motion for a new trial and amendment to the findings of fact. The trial court denied those motions. Admittedly, it would have been the better practice for the trial court in its order denying the motion to explain the misstatement of evidence contained in its memorandum opinion. The trial court, however, made no mention of the melted heater fins or the cause of such melting in its formal findings of fact. Thus the trial court was not required to amend those findings of fact if the statement contained in the memorandum opinion was a misstatement of the evidence which the trial court did not rely on in reaching its decision. If there is a discrepancy between the memorandum opinion and the findings of fact the latter prevails. Rack v. Rack, 142 N.W.2d 754 (N.D.1966); United States v. Cornish, 348 F.2d 175 (9th Cir. 1965). Certainly, in a highly technical case such as this lasting a total of eight days, we cannot expect a trial court’s summary of the evidence contained in its memorandum opinion to be letter perfect. Generally we review the findings of fact labelled as such, and only if they are grossly inadequate will we examine the memorandum opinion for the purpose of determining the facts found by the trial court. Hegge v. Hegge, 236 N.W.2d 910, 914 (N.D.1975). We find no such gross inadequacies in this case.
VI
McGraw-Edison also took issue with the trial court’s imposition of liability on grounds of strict liability and failure to warn. McGraw-Edison’s argument is premised on the contention that the plaintiffs failed to prove a defect in the function or design of the wall heater. This court has addressed the major points of this issue on prior occasions. In Herman v. General Irrigation Co., 247 N.W.2d 472, 476 (N.D.1976), we said:
“Liability based on either strict liability or breach of warranty cannot attach unless there is proof of a defect. . While the strict liability concept relieves the plaintiff of the burden of proving negligence, it does not relieve him of showing a defect in the product.”
See also, Olson v. A. W. Chesterton Co., 256 N.W.2d 530 (N.D.1977).
*802In the Herman opinion, however, we also stated, at page 478:
“But conclusive proof is not necessary to show existence of a defect; it can be done by circumstantial evidence. Langford v. Chrysler Motors Corp., 513 F.2d 1121 (2d Cir. 1975). It is not enough simply to show that an accident or mishap occurred to establish a defect, but the nature of the defect need not be precisely established, especially if a complex product is involved. A defect may be inferred from proof that the product did not perform as intended by the manufacturer. Hursh, American Law of Products Liability 1:11; Lindsay v. McDonnell Douglas Aircraft Corp., 460 F.2d 631 (8th Cir. 1972). Expert testimony is of great value in establishing defectiveness, and here such testimony was presented.” [Emphasis added.]
McGraw argued there was no evidence to support a finding that the heater was dangerous next to drapery material as to show a defect or faulty design. We disagree. The testimony of Bhatt indicated that under certain conditions the heater was capable of producing temperatures sufficient to cause ignition of the drapes. The design of a home heating unit which allows the heater to reach temperatures sufficient to ignite common household goods with which the unit can be expected to come in contact, can be considered an unreasonably dangerous defect. Testimony also indicated an alternate design in the safety switches of the heater would have eliminated or reduced the possibility of such overheating.
“A manufacturer of a chattel owes a duty to the user, although there is no privity of contract between them, to design and manufacture the chattel so as to make it reasonably safe for the use for which it was intended.” Lindenberg v. Folson, 138 N.W.2d 573, Syllabus ¶ 8. (N.D.1973).
See also, Olson v. A. W. Chesterton Co., supra.
We find the evidence sufficient to support the trial court’s imposition of strict liability on the grounds of the malfunction or faulty design of the heater. To prove a product defective, a plaintiff is not required to eliminate with certainty all other possible causes of the accident, but rather to present sufficient evidence to allow the trier of facts to reasonably infer that it was more probable than not that the product was defective. See Daleiden v. Carborundum Co., 438 F.2d 1017 (8th Cir. 1971); 63 Am.Jur.2d Products Liability, § 130.
VII
McGraw-Edison asserted that because evidence produced at trial indicated no one in this case was ever aware of an electric wall heater ever causing a fire such as this, there could be no liability on grounds of failure to warn. In Seibel v. Symons Corp., 221 N.W.2d 50, 54 (N.D.1974), we quoted from our opinion in Lindenberg v. Folson, supra at 582, where we said:
“ . . . a manufacturer supplying a machine has a duty to exercise reasonable care to inform the user of any dangerous condition and character of the machine when put to the use for which it was manufactured and sold, where such danger is known or which reasonably should have been known by the manufacturer in the exercise of ordinary care.”
A product may be considered “defective” so as to put into operation the strict liability doctrine if the manufacturer or seller has reason to anticipate a danger from the use of the product and fails to give an appropriate warning. 63 Am.Jur.2d Products Liability § 131, p. 136.
However, in this case the trial court found as a fact and held that the damage was caused by the “malfunction or faulty design” of the heater, which finding of fact was supported by the evidence. This, in effect, declared the equipment defective so as to put into operation the theory of strict liability without the need of establishing the failure to warn. Because liability can properly be supported in this case by the finding of the malfunction or faulty design of the wall heater, we need not reach the question of whether or not McGraw-Edison knew or should have rea*803sonably known of the possible danger from the defect.
VIII
Finally, McGraw-Edison argued the trial court abused its discretion in assessing costs against McGraw-Edison in the amount of $4,929.28 as expert witness fees for Bhatt. It argued that such costs should not have been awarded when it was shown that Bhatt was an employee of the General Adjustment Bureau which is a firm engaged in investigating and adjusting claims for insurance companies, and in this case, the carrier that paid the loss on Pat’s Motel. In addition, McGraw asserted there is no evidence plaintiff was ever charged for the services of Bhatt and that McGraw-Edison should not be liable for costs not actually incurred.
By way of background, we note that during the course of trial the court imposed a condition upon McGraw-Edison when it sought to introduce the testimony of an expert not listed at the pretrial conference and who remained undisclosed until midnight of the night before he testified. That condition, to which McGraw agreed, was that McGraw-Edison would have to pay for plaintiffs’ expert if plaintiffs felt one was necessary to rebut the testimony of the defendant’s unexpected witness. Plaintiffs subsequently recalled Bhatt to serve as that rebuttal witness. In its conclusions of law the trial court determined that MeGraw-Ed-ison was liable for the plaintiffs’ costs and disbursements of the trial. Counsel for the Schmidts subsequently filed an affidavit of costs and disbursements showing an amount for Bhatt of $5,142.44. Following an objection to the costs filed by McGraw-Edison, the trial court ruled Bhatt served three purposes at trial: (1) as an expert on behalf of the plaintiffs Lawrence and Pat Schmidt; (2) as an expert for Home Insurance Company who was not a party to the action; and (3) to assist plaintiffs’ counsel with technical trial evidence presented in court by the defendant. In view of this determination the court approved total costs in the amount of $4,929.28, including $2,642.44 for expert witness fees for Bhatt. McGraw-Edison’s challenge, then, really goes only to this latter amount.
The determination of the amount of expert witness fees are matters appropriately left to the discretion of the trial court. Peterson v. Hart, 278 N.W.2d 133 (N.D.1979); City of Bismarck v. Thom, 261 N.W.2d 640, 647 (N.D.1977). Evidence adduced at the trial, to the value of services of an expert of Bhatt’s caliber, supported the conclusion that the costs awarded in this case were reasonable. The mere fact Bhatt may have been a salaried employee representing an investigating agency employed by an insurance company involved in this action does not mean costs for his services were not incurred. Certainly his appearance at trial prevented Bhatt from performing other functions during that time. An expert’s bias or place of employment goes to the expert’s credibility and not to his qualifications as an expert once a foundation for that status has been established. Thus, the question of the award of costs and the amount thereof remains in the discretion of the trial court. We find no abuse of discretion in the award of expert witness fees in this case.
The order and judgment of the district court are affirmed.
ERICKSTAD, C. J., and PAULSON and VANDE WALLE, JJ., concur.. In Warner v. Johnson, 213 N.W.2d 895, 897, 898 (N.D.1972), this court said:
“We disapprove of the practice of a trial judge’s uncritically accepting proposed findings, but this unfortunate practice does not erase the ‘clearly erroneous’ rule.” [Underscoring ours.]
Several Federal appellate courts have also criticized this practice. However, with the present strain on the judicial resources we do not disapprove of the practice of having the prevailing party prepare the findings of fact, and believe it is unlikely that the practice will change: At the same time, we note Rule 52(a) also authorizes using the memorandum opinion to serve as findings of fact and conclusions of law. Under the present circumstances this may be the better practice.