Defendant appeals the abatement of plaintiffs’ (PSNH) taxes for the years 1983 through 1986. Following trial on PSNH’s petition for abatement of taxes, the Master (Peter V. Mill-ham, Esq.) made a report in favor of PSNH and recommended an abatement of taxes of approximately $30 million. (PSNH is the owner of 35% of the nuclear plant located in the town of Seabrook. There were, in addition, sixteen other owners of the plant, with lesser interests, when this litigation was commenced. All of the owners are collectively referred to as PSNH.) The Superior Court (Contas, J.) approved the master’s report on October 19, 1988. The defendant (Seabrook) presents three questions for our review:
(1) Does the master’s ruling allowing allocation of indirect construction costs to direct non-taxable construction costs on a dollar-for-dollar basis violate the parties’ pretrial stipulation?
(2) Was it error for the master to allow PSNH’s expert, a property appraiser, to testify concerning the non-operable discount factor used to calculate the power plant’s fair market value?
(3) Are the department of revenue administration’s equalization ratios, adopted by the master to determine the power plant’s assessment value, fairly representative of the general level of assessment of property in the town of Seabrook?
For the reasons stated below; we rule in favor of PSNH on issues two and three, but reverse and remand on the first issue.
This case involves the taxation of PSNH’s nuclear power plant, located in the town of Seabrook. In order to determine the amount of tax money PSNH must contribute to Seabrook’s coffers, Seabrook must perform two tasks. First, it must compute the fair market value (FMV) of PSNH’s taxable property located in Seabrook. Issues number one and two, listed above, concern different aspects of this first *368step. Second, Seabrook must use the FMV of PSNH’s property to calculate the assessment value, the value at which a property is taxed. Issue number three concerns this second step.
I. Effect of Pretrial Stipulation
Before trial, the parties signed a stipulation with the following provision:
“5. The methodology to be employed to determine the fair market value of the Seabrook project, an incomplete nuclear power project under construction, shall be the methodology adopted by the Master in PSNH, et al. v. Seabrook, Docket Nos. E-604-79, et al. in his report dated August 24, 1984, as modified upon remand from the New Hampshire Supreme Court in his Report dated October 24,1986. The parties are agreed that any items of pollution control equipment entitled to an exemption pursuant to RSA 72:12-a shall be considered non-taxable property in the determination of fair market value.”
The master in this case approved the stipulation.
The earlier case (“PSNH, /”) referred to in the stipulation was a tax abatement case similar to this one, but involving earlier tax years. Public Serv. Co. of N.H. v. Town of Seabrook, 126 N.H. 740, 496 A.2d 352 (1985). In both cases the parties agreed that the power plant’s FMV should be determined by adding up all costs incurred by PSNH during its construction, subtracting appropriate deductions, and applying a discount factor. Since the plant was not in operation during the years in question, original cost of construction, less discount and deductions, was considered the best measure of FMV.
To determine FMV and, thus, how much of the plant’s total cost is subject to taxation, Seabrook must deduct from the plant’s total cost all expenditures attributable to non-taxable items. Examples of nontaxable items are: (1) equipment and materials located off-site; (2) equipment and materials located on-site but not yet installed; and (3) exempt pollution control equipment.
Direct costs, such as material and labor, are easily attributable to taxable or non-taxable items, and there is no dispute between PSNH and Seabrook over which direct costs are taxable and which are not. Most indirect costs, however, must be allocated between taxable and non-taxable property, and they are more difficult to attribute to specific items. Engineering, inspection, public relations, administration, and legal expenses are all examples of indirect costs.
*369In PSNH, I, as in this case, a dispute arose over how to tax indirect construction costs. Because these indirect costs are difficult to allocate precisely between taxable and non-taxable direct costs while construction is on-going, Seabrook wanted to treat most indirect costs as taxable until the year construction was complete. At that time, Seabrook would make a careful calculation of which indirect costs were attributable to taxable items and which were not.
PSNH, on the other hand, protested that Seabrook’s method would unfairly result in current taxation of non-taxable items. As a solution, PSNH proposed an allocation of indirect costs between taxable and non-taxable direct costs on a dollar-for-dollar basis. That is, PSNH wished to calculate the percentage of direct costs that were non-taxable and then apply that percentage to the indirect costs to arrive at a fair approximation of non-taxable indirect costs.
The master in the present case was also the master in PSNH, I. In that case, he ruled in favor of Seabrook on the issue of taxing indirect costs. PSNH filed a motion to clarify, again urging the master to make a dollar-for-dollar allocation of indirect costs to non-taxable direct costs. The master denied the request, and PSNH did not appeal this denial.
In the trial of the present case, PSNH again argued before the master that a portion of the indirect costs of the power plant should be allocated to the non-taxable direct costs on a dollar-for-dollar basis. Although before the trial in this case began, Seabrook was aware that PSNH planned to argue the dollar-for-dollar allocation position, Seabrook instructed its expert witness that the Town was not required to litigate the issue of allocation of indirect costs beyond the scope of the master’s decision at the first trial. At the close of this trial, however, the master ruled in favor of PSNH on the allocation issue.
Seabrook now argues on appeal that the master’s ruling violated the pretrial stipulation. Seabrook urges us to interpret the stipulation, quoted above, as precluding a ruling in favor of PSNH on the issue of taxing indirect costs. By ruling in PSNH’s favor on this issue, Seabrook argues, the master changed the methodology used in the first trial to determine the power plant’s FMV, and thus violated the parties’ pretrial stipulation.
PSNH disagrees with Seabrook’s interpretation of the pretrial stipulation. According to PSNH, the phrase “the methodology adopted by the Master in PSNH, et al. v. Seabrook [to determine FMV]” refers only to the general methodology used in PSNH, /; that *370is, calculating total cost, subtracting deductions, and applying a discount factor. It thus does not refer, PSNH argues, to determinations as specific as the allocation of indirect costs. We find the stipulation ambiguous, and therefore remand for a determination of the parties’ intent.
A stipulated agreement is contractual in nature and therefore is governed by contract rules. Kilroe v. Troast, 117 N.H. 598, 600, 376 A.2d 131, 133 (1977). One such rule states that “[a] clause is ambiguous when the contracting parties reasonably differ as to its meaning.” Laconia Rod & Gun Club v. Hartford Acc. & Indemn. Co., 123 N.H. 179, 182, 459 A.2d 249, 251 (1983). In addition,
“The general rule is that whether a contract is clear or ambiguous is a question of law subject to de novo review, and New Hampshire decisions appear to reflect this approach. See, e.g., Restaurant Operators, Inc. v. Jenney, 128 N.H. 708, 519 A.2d 256, 258 (1986). If the contract is deemed ambiguous, then the intention of the parties is a question of fact, id.; MacLeod v. Chalet Susse Int’l, Inc., 119 N.H. 238, 243, 401 A.2d 205, 209 (1979), and the [trial court’s] finding . . . will be set aside only if clearly erroneous.”
In re Navigation Technology Corp., 880 F.2d 1491, 1495 (1st Cir. 1989) (citations omitted).
Because PSNH and Seabrook can reasonably differ as to its meaning, we find the stipulation ambiguous. Its language offers no straightforward answer to our question, “What exactly did the parties mean by the word ‘methodology,’ as used in the stipulation?” The master’s August 24, 1984 report, referred to in the stipulation, describes in detail the methodology the master used to calculate FM^ but the report does not indicate to us whether the stipulation binds the parties to the master’s specific methodology or, rather, to his general methodology.
“While the interpretation of a contract is generally a question of law for the court, when there is a disputed question of fact as to the terms of a contract, it is to be resolved by the trier of fact.” Peabody v. Wentzell, 123 N.H. 416, 418, 462 A.2d 105, 107 (1983); see also Restaurant Operators, Inc. v. Jenney supra; MacLeod v. Chalet Susse Int’l, Inc. supra.
The master did not determine what the parties intended by the term “methodology.” Although he discussed the stipulation’s provisions in the opening paragraph of his report, he did not mention the *371stipulation, or the interpretation problem, in the section of the report concerning allocation of indirect costs. The stipulation is briefly-addressed in the master’s rulings on two of Seabrook’s requests for findings of fact and rulings of law. The requests read:
“14. Except as modified by the Master’s report of 10/24/86, the Master made no further allocation of indirect costs between taxable and non-taxable direct costs. The plaintiffs, subsequent to the filing of the Master’s report of 8/24/84, filed a motion to clarify which among the points, urged the Master further to allocate all indirect costs (primarily engineering) between direct taxable costs and direct non-taxable costs. (See motion to clarify to which are attached Schedules F, G and H of Exhibits 4 and 5 — Stone & Webster reports and also transcript of post-trial hearing held 9/24/84 at pp. 9-14.) The Master denied the motion to clarify insofar as it related to the allocation issue, and his supplemental report dated 9/25/84 was approved by the Superior Court.
“16. The plaintiffs are not entitled to re-open the allocation issue in this trial in the light of par 5 of the pretrial stipulation, on which the defendant had a right to rely, as it did, in preparing and presenting its defense (MacArthur testimony, Tr. vol III pp. 81-88; see Kelly testimony at first trial (Tr. vol II, p. 143)).”
The master denied Seabrook’s request number 16 without comment, and granted number 14 with the following qualifying remark:
“[Granted] insofar as the Master cannot determine what allocation he made of indirect costs in the earlier decision. The Master does not consider that he is prevented from making an allocation at this point where both parties’ appraisers have agreed that such an allocation is proper . . . .”
In ruling on the stipulation, however, the master never addressed the question of the parties’ intent; therefore, the meaning PSNH and Seabrook ascribed to the term “methodology” in their stipulation is unknown. We must remand the issue of intent to the master for a determination of this question of fact, because we cannot properly make such a determination.
*372II. Plaintiff’s Expert Witness Testimony Concerning the Discount Factor
To determine this power plant’s FM^ one must add together the plant’s construction costs and then apply to that figure a discount factor. While the parties agree that a discount factor should be applied, they disagree over the size of the discount factor and, more importantly, over its nature. Seabrook argues that the sum of construction costs should be discounted only to reflect an investor’s unwillingness to pay full value now for income to be received in the future. PSNH contends that, in addition, the sum of construction costs should be further reduced to account for the risk that the plant’s operation would be delayed or that the plant would never go on-line.
PSNH’s expert witness, John P. Kelly, is an appraiser and an acknowledged expert in the valuation of regulated public utility property, including nuclear power plants. He calculated a discount factor for PSNH’s plant and testified that, in part, the discount factor quantifies the cost of money between the time of construction and the time of operation. In other words, Kelly accounted for an investor’s unwillingness to pay a full dollar now for a dollar to be received at a future date.
In addition, Kelly testified that he analyzed the histories of other American nuclear power plants and quantified the risks of delay and cancellation from this data by using probability theory. Last, he testified that he examined the problems particular to PSNH’s power plant and from that analysis calculated a further quantification of the risks of delay and cancellation. He based his discount factor on these elements.
Seabrook objected to the admission of Kelly’s testimony at trial, and renews that objection on appeal. Kelly was not competent to testify concerning the risks of delay and non-operation, Seabrook argues, because Kelly is an expert appraiser, not an expert actuary. Seabrook maintains that the quantification of such risks is part of the actuarial science and requires the skill of a trained actuary.
Seabrook’s chief expert witness, David MacArthur, is also not an actuary. His discount factors did not reflect any risk of delay or non-operation. Instead, MacArthur simply discounted the FMV to present value to account for an investor’s unwillingness to pay a full dollar now for a dollar to be received in the future. Lawrence Patz, a professional actuary, testified for Seabrook and agreed with MacArthur’s position:
*373“I believe some discounting should take place as to the present value of future profits at the yield rate. But beyond that, there is no data or no mathematical justification for any discounting at all.”
On cross-examination, however, Patz testified that a prospective buyer of a nuclear power plant would likely use Kelly’s discount factors to arrive at a fair price:
“A. I think a prudent buyer would try to do what [Kelly’s] appraisal reports have done in order to lower the price. That is good business practice. It is not good mathematical theory.
Q. So, a buyer would go through the processes as set forth, that are set forth in the appraisal?
A. That or a similar approach, yes.”
We think it was reasonable for the master to consider Kelly’s testimony, because FMV must, by definition, be determined according to the factors a willing buyer and a willing seller would apply on an open market. See BALLENTINE’S Law DICTIONARY 452 (3d edition 1969). Moreover, it is within the province of the finder of fact to weigh the testimony of witnesses. See Albee v. Wolfeboro Railroad Co., 126 N.H. 176, 179, 489 A.2d 148, 150 (1985).
In the end, the discount factors the master applied to the plant’s FMV were higher than Seabrook’s factors, but lower than PSNH’s. In his report, the master did not appear to accept Kelly’s testimony concerning other American nuclear power plants and probability theory. Instead, the master examined the troublesome history of the Seabrook plant itself and arrived at discount factors using his own judgment.
We are not persuaded that the master committed error when he admitted Kelly’s testimony concerning quantification of the power plant’s risk of delay and cancellation. Seabrook argues that Kelly was not qualified to testify on an actuarial issue; however, whether an expert is properly qualified to testify on a subject is a determination within the sound discretion of the trial court. Dowling v. L.H. Shattuck, Inc., 91 N.H. 234, 236, 17 A.2d 529, 532 (1941). Any appraisal of a power plant under construction must necessarily include the quantification of risks of delay and cancellation in order to determine the plant’s true FMV. Since Kelly had ample experience appraising both completed and uncompleted power plants and ap*374peared to have a reasonable basis for making his calculations, the master did not abuse his discretion in admitting Kelly’s testimony.
III. Equalization Ratios
The purpose of the trial below was to determine the power plant’s assessed value; that is, the value at which the plant should be taxed. A property’s assessed value is not necessarily its FMV but is instead usually a smaller proportion of FMV. In order to calculate what proportion of FMV the plant’s assessed value should be, we must examine the relationship between the assessed value and FMV of all other property in the town of Seabrook. This is because the ratio of a property’s assessed value to FMV must not be any higher than the ratio of assessed values to FMV’s throughout the rest of the town. In this way, no property’s assessed value is disproportionately higher in relation to its true value than is the case as to other property in the town. Berthiaume v. City of Nashua, 118 N.H. 646, 647, 392 A.2d 143, 144 (1978).
In 1982, all property in the town of Seabrook was individually appraised. In that year, the properties were assessed at the appraisal value, and thus assessed value equaled FMV. In the years following 1982, however, no municipality-wide appraisals were performed. Although inflation caused the FMV’s to rise, Seabrook had no reliable method for determining the increased FMV of individual properties without an appraisal. Therefore, Seabrook simply kept the properties’ assessed values at the 1982 level. As inflation pushed the FMV’s higher each year, the assessed values became, on average, a smaller and smaller proportion of the FMV’s.
For most property owners, the difference between the property’s assessed value and its higher FMV was of no concern, since the owners were still taxed at the lower 1982 level. However, taxation for some property owners, such as PSNH, was complicated when the FMV of their property increased dramatically due not just to inflation, but also to on-going construction and “improvement” of the property. For such property, the 1982 level would no longer bear a purely inflationary relation to FMV.
Since the proper assessed value of a property, in inflationary times, is simply its FMV minus the effect of inflation since the last appraisal, PSNH’s power plant should also be assessed at a value equal to its FMV minus the effect of inflation since the 1982 appraisal. Seabrook attempted to follow this formula to determine the power plant’s assessed value. Although the record does not unam*375biguously indicate the method Seabrook originally used to determine the power plant’s assessed value, it appears to be the following. First, Seabrook calculated the plant’s FMV. Second, it used the Handy-Whitman Index, a public utility inflation index, to figure the effect of inflation. Subtracting the effect of inflation from the FMV Seabrook arrived at an assessed value for PSNH’s power plant. Sea-brook’s original calculation produced the following ratios: 100% for 1983, 100% for 1984, 97.9% for 1985, and 93.8% for 1986. The town applied these ratios respectively to the power plant’s FMV to arrive at assessed values in the years 1983 through 1986. PSNH was taxed at these assessed values.
The Handy-Whitman Index, used by Seabrook, does not take into account the relation between assessed value and FMV (the inflationary effect) for all other property in the town. As a result, the ratio of assessed value to FMV for the power plant was much higher than that for all other property in Seabrook. PSNH sued for tax abatement, claiming that Seabrook forced it to pay a disproportionately high share of Seabrook’s property taxes. The burden was properly placed on PSNH to prove disproportionality. Berthiaume, 118 N.H. at 647, 392 A.2d at 144.
Prior to trial in this case, but after PSNH was assessed according to the ratios listed above, we decided PSNH, I. MacArthur, the town’s tax assessor, realized that the original assessments of PSNH were no longer accurate under PSNH, I, and thus revised them in preparation for this trial. The revisions mostly concerned the calculation and taxability of “AFUDC” accounts (“Allowance for Funds Used During Construction”). MacArthur testified:
“I have revised all appraisals ... in accordance with the decision that came out and had gone through the Supreme Court. ... I don’t have any previous appraisals [with me in court] because I have re-written them all. As far as I am concerned, they are null and void .... We are not dealing with those appraisals.”
The new ratios calculated by MacArthur and presented by him at trial are 98.6% for 1983, 96.6% for 1984, 93.9% for 1985, and 91.9% for 1986. He testified that he again used the Handy-Whitman Index to determine the effect of inflation on the power plant’s FMV.
The master in the trial below rejected Seabrook’s original assessments of PSNH, based on the originally calculated ratios, and refused to use MacArthur’s revised ratios to determine new assessments. Instead, the master used the department of revenue *376administration’s (DRA) equalization ratios to arrive at PSNH’s assessed values. An equalization ratio (ER) is a device used by the DRA to figure the total FMV of all the property in a town. The ER represents the average ratio of assessed value to FMV in a town and thus is a measure of inflation for the whole town. It is calculated in the following manner. First, the DRA determines which properties in a town were sold in any given year. For the sales transactions that were at “arm’s-length,” the sale price is assumed to be the property’s FMV. Second, the DRA takes the assessed values of the properties sold at “arm’s-length” and calculates the ratio of assessed value to FMV for each property. Third, the median ratio is chosen as the town’s ER. The median ratio is the middle ratio when the ratios are arranged from lowest to highest.
Seabrook’s ER, as calculated by the DRA, represents the average ratio of assessed value to FMV for the town in any given year, and thus is a measure of inflation. The ER for 1983 was 88%, for 1984 it was 72%, for 1985 it was 63%, and for 1986 it was 47%. The master applied the ER to the power plant’s FMV (i.e., multiplied the ER by the FMV) to arrive at the plant’s proper assessed value. The DRA’s ER’s for the years 1983 through 1986 are lower than MacArthur’s ratios, and thus result in lower assessed values for PSNH’s power plant.
Seabrook, on appeal to this court, no longer argues that its original assessments of PSNH should stand (MacArthur declared them “null and void,” and Seabrook’s brief does not advocate their use). Instead, Seabrook urges us to adopt either MacArthur’s revised ratios or Seabrook’s “aggregate ratios,” explained below. Before discussing the merits of Seabrook’s alternative ratio formulas, we will address Seabrook’s complaint that the DRA’s ER’s, which the master below adopted, are unconstitutional.
Seabrook objects to the use of the DRA’s ER’s to calculate the plant’s proper assessed value, because it claims the ER’s do not fairly represent the property in Seabrook and thus violate the State’s constitutional requirement of proportionality and uniformity in taxation. See N.H. CONST. pt. I, art. 12, pt. II, art. 5; Bemis &c. Bag Co. v. Claremont, 98 N.H. 446, 450, 102 A.2d 512, 515 (1954). More specifically, Seabrook points out that the ratio of the plant’s assessed value to its FMV was not included in the DRA’s calculation of each ER. However, the fact that the DRA did not take into account public utility property in its ER calculation does not render that calculation invalid or unconstitutional. It is not necessary in attempting *377to prove disproportionality that PSNH use only comparable properties. Berthiaume, 118 N.H. at 648, 392 A.2d at 145. Similarly, we hold that it is not necessary in determining a proportional assessment here that PSNH use only comparable properties.
MacArthur’s revised assessments, described above, should not be adopted by this court for two reasons. First, the effect of inflation is calculated by use of the Handy-Whitman Index. This index, though a widely recognized indicator of public utility property inflation, is an inappropriate tool for use in a tax abatement case. The purpose of this case is to ensure that PSNH is not taxed at a level that is disproportionately higher in relation to its true value than is the case as to other property in Seabrook. Berthiaume, 118 N.H. at 647, 392 A.2d at 144. Thus it is essential that we compare the power plant’s ratio of assessed value to FMV with such ratios for the remainder of the property taxed in Seabrook. The Handy-Whitman Index does not take into account the ratios of assessed value to FMV for property other than public utility property, and therefore it is not appropriate to use it to determine the power plant’s assessed value.
Second, the use of the Handy-Whitman Index is inappropriate in the calculation of PSNH’s assessed value because it results in two ratios for the town of Seabrook in each tax year: one higher ratio for public utility property (MacArthur’s ratio), and one lower ratio for the rest of the property in Seabrook (the DRA’s ER, which Sea-brook agrees is accurate for non-public utility property). We have previously held that such a practice is invalid.
“It is impermissible to maintain a class of real estate that is assessed at a higher level than other real estate whether that class consists of one parcel or half the town. It is therefore irrelevant that all assessments within one such class may be uniform. Widespread disproportionality is no defense.”
Appeal of Town of Sunapee, 126 N.H. 214, 219, 489 A.2d 153, 156 (1985). Again, our constitution mandates that all taxpayers in a town be assessed at the same proportion of FMV. Berthiaume supra.
Seabrook argues in the alternative for the use of an “aggregate ratio,” and asserts that this single ratio best represents the relationship between assessed values and FMV’s for all the property in Sea-brook. The “aggregate ratio” as advocated by Seabrook is calculated by dividing the net locally assessed value of all the property in the town, as determined by Seabrook in its original assessment, by the total FMV of the property in the town, as determined by the DRA.
*378Seabrook argues that this ratio is more appropriate than the DRA’s ER, because the ER is calculated using a median ratio. A median ratio places the same weight on each property used, no matter how much the property is worth. According to Seabrook, the value of parcels of property should have more impact on a ratio than simply the numbers of parcels. We disagree. The median ratio is in essence the “one taxpayer, one vote” method, and prevents one unusually valuable property from dominating the ratio and, thus, the other taxpayers’ assessments;
Moreover, Seabrook’s “aggregate ratio” is identical to the “composite ratio” advocated by the Towns of Bow, Newington, and Seabrook in Appeals of Bow & a., 133 N.H. 194, 201, 202, 575 A.2d 1301, 1306 (1990). We rejected the “composite ratio” in that case because one of the ratio formula’s components was the subject of the case. Id. at 201, 202, 575 A.2d at 1306. We reject the “composite/ aggregate ratio” in this case for the same reason. The net locally assessed value of the property in Seabrook, one of the formula’s components, is in part made up of the power plant’s assessed value, the subject of this case, as originally calculated by Seabrook. We cannot properly consider in this case any ratio that is calculated by using an assessed value assigned to the power plant. “The issue is whether the plaintiff’s tax is greater than it should be with respect to the taxes of other property owners in the taxing district.” Bemis &c. Bag Co. v. Claremont, 98 N.H. 446, 449, 102 A.2d 512, 514-15 (1954) (emphasis added).
We hold that the DRA’s ER’s for 1983 through 1986 are the appropriate ratios to use to calculate PSNH’s proper assessed value. PSNH introduced ample evidence supporting the statistical validity of the DRA’s ER’s, as well as a proportionality study of an independent expert based on a survey of a fair cross section of the town’s real estate. See Milford Props., Inc. v. Town of Milford, 119 N.H. 165, 168, 400 A.2d 41, 43 (1979). As we stated above, the DRA’s ER’s were not rendered invalid simply because they did not reflect the ratio of the assessed value of public utility property to its FMV. PSNH has thus met its “burden of showing that the assessment placed on the subject property was disproportionately higher in relation to its true value than was the case as to the other property in the [town].” Berthiaume, 118 N.H. at 676, 392 A.2d at 144.
Affirmed in part; reversed in part; remanded.
HORTON, J., did not sit; the others concurred.