dissenting:
My colleagues in the majority affirm the trial judge’s decision, which for the most part sustains the valuation of the subject property by the Department of Finance and Revenue (DFR), even though
1. on January 1, 1988, the date on which the assessor, Troy Davis, assessed the value of the property for the year 1989 at $38,760,980, Davis was unaware of the fact that the property had been sold at arm’s length, less than thirteen months earlier, for $22,000,000;
2. the trial judge excluded evidence showing that in an assessment of the property conducted for the second half of the year 1988, dated December 31, 1987, the very same assessor had valued the property at $23,731,600; and
3. the judge accepted DFR’s valuation over that proposed by the taxpayer’s expert, Michelle Saad, who, unlike Mr. Davis, did take into consideration both the 1986 sales price and the December 31, 1987 valuation.
In my opinion, the deference that courts ordinarily accord to agency expertise is sorely tested by this record. Indeed, I am persuaded that the rather dramatic omissions and inconsistencies on the part of the DFR have so impaired the reliability of the agency’s valuation that the process must begin anew. Accordingly, I would reverse the judgment.
I.
THE DECEMBER 1986 SALE
A. The assessor’s error.
On December 12, 1986, the taxpayer purchased the subject property from Woodward & Lothrop for $22,000,000. The taxpayer presented testimony showing that the sale was at arm’s length, and there is no evidence to the contrary. Although, as the trial judge found, the sale was duly recorded, the assessor was unaware of the transaction when he made the 1989 assessment less than thirteen months after the sale. The taxpayer contends that Mr. Davis’ failure to ascertain and incorporate into his calculus a critical fact about the property that he was supposed to be assessing deprived his valuation of all or most of its probative force. I find it difficult to disagree.
It is surely beyond dispute that the assessor was obliged by law to become aware of, and to consider, a recent arm’s length sale. Our real property assessment statute provides that the Mayor (and therefore his agent, the assessor) “shall take into account any factor which might have a bearing on the market value of the real property .... ” D.C.Code § 47-820(a) (1997). An applicable regulation tracks the statute. 9 DCMR § 307.1 (1996). “Sales which represent arm’s length transactions between buyer and seller shall be used in analyzing market values.” Id. § 307.3(a). Moreover, the Recorder of Deeds is required by regulation to transmit the deed recordation tax return to DFR within five days, 9 DCMR § 312.1, and DFR thus receives notice of a duly recorded sale.1
The “estimated market value” of assessed property is defined in D.C.Code § 47-802(4) (1997) and quoted by my colleagues. “[F]air market value is the price property will bring when offered for sale by a seller who desires but is not obliged to sell and bought by a buyer under no necessity of purchasing.” Board of Supervisors of Fairfax County v. Donatelli & Klein, Inc., 228 Va. 620, 325 S.E.2d 342, 345 (Va.1985) (citations omitted). *973As a matter of simple common sense, a recent arm’s length sale must be a very significant factor, if not the most significant factor, in determining fair market value. The authorities are uniformly to the same effect.
Rule 1-5 of the Uniform Standards of Professional Appraisal Practice (USPAP) (1992) provides in pertinent part:
In developing a real property appraisal, an appraiser must:
(b) consider and analyze any prior sales of the property being appraised that occurred within the following time periods:
(i) one year for one-to-four family residential property; and
(ii) three years for all other [including commercial] ‘property types;
Comment: Departure from binding requirements (a) through (c) is not permitted.
(Emphasis added.) In failing to ascertain that the property had recently been sold, and in failing to include the sales price in his valuation calculus, the assessor thus departed from a fundamental precept of appraisal practice.2
The significance of a recent arm’s length sale has also been recognized by the courts. “As is said in 84 C.J.S. Taxation § 576-c, pp. 1147, 1148, ‘evidence of the purchase price of the assessed property, while not conclusive, is to be accorded substantial weight on the issue of fair market value.’ ” American Viscose Corp. v. City of Roanoke, 205 Va. 192, 135 S.E.2d 795, 798 (Va.1964); accord, Arlington County Bd. v. Ginsberg, 228 Va. 633, 325 S.E.2d 348, 352 (Va.1985) (“[wjhere there has been a recent sale of property, of course, such sale should be considered”); Donatelli & Klein, supra, 325 S.E.2d at 345; Belk Dep’t Stores v. Taylor, 259 S.C. 174, 191 S.E.2d 144, 146 (S.C.1972) (“[w]hile other circumstances may affect the question, the cost of an article is ordinarily some evidence of its value for tax purposes”); People ex rel. 83rd East End Corp. v. Miller, 265 A.D. 14, 37 N.Y.S.2d 677, 678 (N.Y.App.Div. 1st Dep’t 1942) (per curiam). As the court stated in United States v. 428.02 Acres of Land, Etc., 687 F.2d 266, 271 (8th Cir.1982), “a recent sale of the very property to be condemned, provided that the sale is an unforced, arm’s-length transaction, is extremely probative evidence of fair market value.” See also W.R. Habeeb, Annotation, Admissibility, in Eminent Domain Proceeding, of Evidence as to Price Paid for Condemned Real Property on Sale Prior to the Proceeding, 55 A.L.R.2d 791 (1957).3
B. The relevance of the assessor’s error.
My colleagues in the majority do not contest the taxpayer’s assertion that the assessor erred in failing to learn of and consider the 1986 sale of the property. Rather, they take the position, as did the trial judge, that the assessor’s omission is irrelevant to the result. They base this theory on the fact that, although the sale was effected on December 13, 1986, the purchase price was established on November 13, 1983 in an op*974tion agreement between Woodward & Loth-rop and the taxpayer. Ryland Mitchell, an appraiser who testified as an expert on behalf of the District, stated that he attached little significance to the sales price because of the lapse of time since the end of 1983.
I do not deny — no reasonable person could — that the date of the option agreement is a relevant factor and may tend in some measure to alleviate the consequences of Davis’ omission. In my opinion, however, the trial judge went too far, and so do my colleagues, in altogether disregarding the assessor’s error. The price of land tends to appreciate, and one cannot reasonably suppose that, in 1983, the sophisticated seller and buyer were blind to the possibility that the value of the property might increase during the option period, or that they failed to take that prospect into consideration in negotiating the option price. In fact, there was testimony that, at the time of the option agreement, the parties did not intend to close for at least two years, that they never contemplated that the taxpayer would pay $22,-000,000 in 1983, and that the price was designed for a future date.
In any event, at the time that the assessor made the 1989 assessment, he did not know anything either about the 1986 purchase price or about the 1983 option agreement on which that price was based. If Mr. Davis had known the facts which it was his duty to know, he could have inquired into the circumstances, and he would then have been in a position to make an objective determination as to the appropriate weight to be accorded to a December 1986 sale pursuant to a November 1983 option under the circumstances of this case.4 The record as it stands, on the other hand, presents us instead with a self-serving “after-the-fact” declaration that the DFR’s valuation for 1989 was correct notwithstanding the assessor’s embarrassing error.
Moreover, there is compelling independent evidence that the December 1986 purchase price was not out of line with the market value of the property. In December 1987, a year after the sale, DFR found the fair market value of the property, for purposes of its assessment for the second half of 1988, to be $23,731,600. The assessor who made this valuation was Troy Davis. If Mr. Davis had been aware of the 1986 sale, and if he had known that the purchase price was only slightly lower than his own estimate of the value of the property a year later, he might well have accorded the purchase price some significance. At the very least, it would have been logical for him to do so.
Finally, even if — and it is a very big if— the four years that elapsed between November 1983 and January 1, 1988 supported the trial judge’s view that the assessor’s error in failing to consider the sale did not affect his valuation, that error would still be significant. In determining the fair market value of the property, the judge was faced with conflicting estimates by the expert witnesses for the District and for the taxpayer. The failure of Mr. Davis to learn of, and to consider, a critically important fact about the property that he was assessing has obvious relevance to the weight to be given to his testimony.
II.
THE TAX YEAR 1988 SECOND HALF ASSESSMENT
The taxpayer proffered at trial that on December 31, 1987, the property was assessed by Troy Davis at $23,731,600. When the taxpayer’s attorney sought to question Mr. Davis on this subject, however, the evidence was excluded:
*975Mr. Hahn:5 Now isn’t it true that you also did the second half assessment for tax year 1988?
Mr. Davis: That’s correct.
Mr. Hahn: And will you tell the court what value you placed on the land for the second half?
Mr. Ferguson:6 Objection, your Honor _What was done for prior years, assessment years is irrelevant to the case here ....
Mr. Hahn: Your honor, the valuation date for the second half of [1988] is December 31, 1987, which is one day from January 1, 1988. So it’s relevant to show the valuation one day apart in this piece of land.
[T]o have a difference of 63 percent in the value of this land between one year and the next where the valuation dates are only a day apart requires some explanation from the assessor ....
After the attorneys had argued the issue, the taxpayer’s counsel asked Mr. Davis whether he agreed that the 1989 assessment represented a 63 percent increase over the prior year. Counsel for the District objected, and the judge sustained the objection:
The Court: Now if you’ve got something more which, you know, you haven’t introduced into evidence when it comes in maybe I’ll be more receptive, but just to say because it’s a dramatic jump and nothing else into evidence, [then] that must mean that something’s wrong, I can’t buy that. So I’m not going to let you get that in.
(Emphasis added.) The judge ruled, in other words, that the 1988 assessment was not admissible because that assessment did not conclusively prove that “something’s wrong” with the 1989 assessment.
In my opinion, the judge applied an incorrect legal standard. Admissibility is not to be equated with conclusiveness. “[I]f the evidence offered conduces in any reasonable degree to establish the probability or improbability of the fact in controversy, it should go to the jury.” Home Ins. Co. v. Weide, 78 U.S. (11 Wall.) 438, 440, 20 L.Ed. 197 (1870) (emphasis added); see also Martin v. United States, 606 A.2d 120, 128 (D.C.1991) (quoting Weide). Professor McCormick has articulated the point effectively:
It is enough if the item could reasonably show that a fact is slightly more probable than it would appear without that evidence. Even after the probative force of the evidence is spent, the proposition for which it is offered still can seem quite improbable. Thus, the common objection that the inference for which the fact is offered “does not necessarily follow” is untenable. It poses a standard of conclusiveness that very few single items of circumstantial evidence ever could meet. A brick is not a wall.
Edward W. Cleary, et al., McCormick on Evidence § 185, at 542-43 (3d ed.1984); see also Street v. United States, 602 A.2d 141, 143 (D.C.1992) (Steadman, J.) (quoting Professor McCormick); Martin, supra, 606 A.2d at 128-29 (same).7
The 1988 assessment surely had some relevance to the issue before the court. By excluding that assessment from evidence, the judge deprived the taxpayer of the opportunity to impeach the assessor’s valuation of the property for 1989 with a prior and potentially inconsistent assessment issued only one day earlier. It is possible that Davis could have satisfactorily explained the overnight 63 percent increase. It is also possible, however, that he could not.
*976Moreover, as we have noted, the evidence in this case required the trial judge to make a determination as to whether it was the District’s witnesses or Ms. Saad who had presented the more persuasive valuation.8 If, as the proffered evidence suggests, Troy Davis’ assessment was inconsistent to a dramatic degree with his own valuation of the same property only a day before, the taxpayer was entitled to confront Davis with the prior valuation and to ask the judge to draw an appropriate inference if Davis could not satisfactorily explain the apparent inconsistency. The trial judge erred in precluding such impeachment.9
III.
CONCLUSION
I recognize that this appeal turns in substantial part on the trial judge’s factual findings. The judge was present when the witnesses testified, while this court is limited to a paper transcript. Our review is therefore deferential, and we cannot second-guess the judge’s determinations of fact. The judge is entitled, for example, to credit Mr. Davis over Ms. Saad, even if the former has been inconsistent and significantly impeached. This is particularly true where, as here, Mr. Davis’ valuation was substantially corroborated by Ryland Mitchell, the District’s appraisal expert.
But “[findings of fact which result from a misapprehension as to the applicable law ... lose the insulation of the ‘clearly erroneous’ rule.” In re Application of L.L., 653 A.2d 873, 880 (D.C.1995) (citing Murphy v. McCloud, 650 A.2d 202, 210 (D.C.1994)). In my opinion, the trial judge’s critical findings in this case were predicated upon, and induced by, incorrect legal standards. Specifically, I believe that the judge failed to apply legal principles set forth in our statute and regulations, in judicial decisions from other jurisdictions, and (indirectly) in accepted appraisal standards, all of which required him to give substantially greater consideration than he did to the assessor’s failure to incorporate the 1986 purchase price into his calculus.10 The judge also erred, in my opinion, in excluding from evidence the December 31, 1987 assessment for the second half of the year 1988, and in precluding counsel for the taxpayer from cross-examining Davis on the apparent inconsistency between his two valuations issued one day apart. If the judge had recognized what I believe to be the significance of the December 1986 sale and the December 1987 assessment, especially when those two factors are considered in combination, and if the evidence had been developed accordingly, the judge’s ultimate decision might well have been significantly different.
At bottom, my differences with the majority may be based in substantial part on our respective perceptions as to how far courts must go in deferring to supposed agency expertise. In the present case, the assessor’s failure to consider, or even to learn of, the 1986 sale, together with the 63 percent increase in valuation from December 31,1987 to January 1, 1988, would surely undermine the confidence of many, if not most, reasonable people in DFR’s judgment. Agencies are supposed to act rationally and consistently. A taxpayer’s obligation to the revenue man should not be based on a bureaucrat’s ea-*977price. In my opinion, the present record has a certain Aliee-in-Wonderland quality and cries out for more rigorous judicial scrutiny of the agency’s actions, and of the trial judge’s countenancing of those actions, than my colleagues believe to be necessary.
Justice should not only be done, but should seem to be done. I would therefore reverse the judgment, remand the case, and direct the trial judge to require DFR to begin all over again. Because my colleagues view the case differently, I respectfully dissent.
. Cf. 1827 M St., Inc. v. District of Columbia, 537 A.2d 1078, 1083 (D.C.1988) ("The filing of the application [for historic designation] puts the Mayor and all his subordinates, including the tax assessor, on notice of its contents. Consequently, the assessor must take into account the effect that such an application, if granted, may have on the estimated market value of the property.”).
. The trial judge stated in his order that "it may be unsound appraisal practice not to consider the sale of the subject property." Nevertheless, the judge gave "little or no weight” to Rule 1-5 because the taxpayer introduced into evidence the 1992 edition of USPAP, while the years in question were 1988 and 1989. The judge also opined that "these standards were established with regard to appraisers and not assessors."
I respectfully disagree with the judge's reasoning. There is no evidence that the 1992 standards were different in this respect from the standards previously in effect; indeed, the pre-1992 authorities are consistent with the 1992 edition of USPAP. In presenting to the court the most recent edition of the standards, the taxpayer’s counsel simply followed the conventional practice of citing the current edition of publications such as, e.g., the District of Columbia Code. Absent some showing that there had been a recent material change in the standards, the judge’s refusal to consider the USPAP was, in my opinion, unwarranted.
I am also unable to agree with the trial judge's differentiation between appraisers and assessors. Factors relevant to the fair market value of the property are necessarily the same, whether that value is being estimated by an appraiser or by an assessor.
. Condemnation proceedings, like tax assessment cases, turn on the estimated fair market value of the property in question.
. In fact, Mr. Davis admitted that, if he had known about the 1986 sale, he would have considered it in making his assessment. He testified that he would have "weigh[ed] the time frame in which the price was negotiated,” that he would have "looked at the sale price in the context of the overall market surrounding the property,” and that he "would have made a judgment based on all the pertinent data involved.” In response to a question in which he was asked to assume that the December 1986 sale was between a willing buyer and a willing seller, at arm’s length, for $22,000,000, Mr. Davis stated that he would have considered that sale more important to the value of the property than the sales prices *975of comparable properties which he did include in his calculus.
. Gilbert Hahn, Esq., counsel for the taxpayer.
. Joseph F. Ferguson, Esq., Assistant Corporation Counsel, who was representing the District of Columbia.
. According to the majority, "the taxpayer has not come forward with any evidence to suggest why the 1988, second half, assessment should be deserving of greater weight than the tax year 1989 assessment, whose [sic] methodology the trial court chiefly accepted .’’In my opinion, the admissibility of the 1988 assessment cannot turn on whether that assessment deserves "greater weight.” The evidence was admissible because ’ it made it more likely that Davis was wrong and the taxpayer’s expert was right.
. Of course, the judge was not bound by the opinions of any of the witnesses. He could properly conclude, for example, that the true value of the property lay somewhere between Davis’ assessment and Ms. Saad’s estimate.
. In the next-to-last paragraph of Part II.B of the opinion of the court, the majority says that the taxpayer has not "come forward with evidence” about the 1988 assessment. My colleagues also rely on representations which have been made to this court in regard to that assessment. In my opinion, it is unfair to fault the taxpayer for the lack of evidence, and to consider the District’s representations to this court, when the taxpayer was prevented by the trial judge from developing a factual record.
The majority also cites Wolf v. District of Columbia, 609 A.2d 672, 675 n. 2 (D.C.1992) and Wolf v. District of Columbia, 611 A.2d 44, 51 (D.C.1992) in support of the trial judge’s exclusion of the 1988 assessment. Neither of the Wolf decisions is in point. In each, the court held that an assessment is not rendered erroneous simply by the fact that it substantially exceeds the previous year's assessment. Neither decision addresses the question whether a prior year's lower assessment by the same assessor is admissible.
.In this connection, see also note 2, supra.