Cutter, J.
Wannacomet Water Company (see St. 1880, c. 27; St. 1908, c. 548; St. 1928, c. 197) serves over 1,800 customers in Nantucket, of whom about 600 are seasonal. Its common stock is owned by Greenwich Water System, Inc.
Wannacomet filed new rate schedules to become effective July 1, 1961. The Department of Public Utilities suspended their operation and initiated an investigation as to their propriety. After hearings in April and May, 1961, interim rates were put into effect to afford Wannacomet some relief, pending experience with the use of recently installed meters. Wannacomet introduced further evidence at resumed hearings in 1962.
On June 29, 1962, the department disallowed the rate schedule. The department stated that Wannacomet’s depreciation reserve was inadequate by $76,000 and directed the transfer of this amount from earned surplus, thereby reducing Wannacomet’s rate base. The department also found “that a rate of return of 6.5 per cent will be adequate” to meet expenses and interest charges, and to provide a return “calculated to allow ... in excess of 8 per [455]*455cent on the equity portion of the rate base.” Wannacomet had sought a higher return. The department’s order was designed to produce net operating income of $45,185 rather than of $55,932, as sought by Wannacomet.
Wannacomet appealed (G. L. c. 25, § 5, as amended through St. 1956, c. 190), among other matters, objecting on the grounds of alleged confiscation and lack of substantial supporting evidence to the 6.5% rate of return allowed, as well as to the downward adjustment of rate base because of the increase in the depreciation reserve. The town appealed principally on the ground that the investment by Wannacomet in installing meters was imprudent to the extent of $117,000. The single justice, at the request of the parties, reserved the appeals upon a consolidated record for the determination of the full court.
The Town’s Appeal.
In 1958, the department ordered Wannacomet to install meters for not less than seven per cent of its customers each year. Wannacomet, with the department’s prior informal approval, decided (for reasons of economy as well as of public relations and because costs were constantly rising) to do this work at once, rather than over several years. The installation was completed in the summer of 1961, for $184,764,2 and was paid for by short term borrowing, without interest, from a parent company.
The town contends that the meters were placed in concrete vaults under the public way or sidewalk area at an unnecessary installation cost of $73 each, instead of a cost of about $8 each for installation inside the houses served. On this issue, the department’s decision points out that installation of the “meters outside of buildings was made after careful consideration of the particular problems” on the island; that the program represented a “considered judgment by men of long experience ’ ’; and that, although [456]*456“inside installation in some instances would have been feasible,” in the light of “the high proportion of seasonal” customers and houses “without basements, the record does not disclose that these instances are substantial.” The department concluded “that the long run advantages of efficiency of meter reading and of safety of the meter . . . will outweigh . . . [the] added investment cost.”
Wannacomet’s president gave testimony about the considerations that led to the outside meter installations. In Nantucket there are numerous seasonal customers. In houses closed during the winter, water left in an inside meter will freeze and break the meter chamber. To protect such a meter, it is necessary to take the meter out and blow out the water. Wannacomet is unwilling to leave this work to the owner’s plumber, because it wishes to take care of its own property. Leakage from a frozen meter is less likely to cause damage if the meter is outside the house rather than inside. Sixty-five per cent of Nantucket houses have no basements. In such houses, in order to obtain a fair measurement, the meter must be installed at a point outside the first take-off of the main service line. The meter must be at an accessible point where it can be read. Outside meters can be read more rapidly and inexpensively than those installed inside buildings.
Even if total or partial inside installation of meters would have been practicable and cheaper than outside meters, the department could properly conclude that many such installations would be inconvenient to customers and to Wan-nacomet alike. Customers with no cellars could well regard the presence of a meter in their living quarters as unsuitable. The department could give weight to the difficulties which Wannacomet might encounter in reading inside meters, particularly in the closed houses of seasonal customers. Eemoval and blowing out of these meters would have involved some recurrent annual expense and nuisance to all concerned. The company reasonably was reluctant to entrust protection of its meters to its customers’ plumbers. Cheaper cellar installations of inside meters would have been possible only in those houses with cellars. The [457]*457department was warranted in relying upon the experience and judgment of Wannacomet’s management. We assume that the department would not have been bound to follow Wannacomet’s decision if that decision had been shown to be plainly unreasonable. See New England Tel. & Tel. Co. v. Department of Pub. Util. 327 Mass. 81, 90-92. See also New England Tel. & Tel. Co. v. Department of Pub. Util. 262 Mass. 137,146. The company seems to have proceeded cautiously and after consultation with the department and its own counsel. Substantial testimony justified the department’s conclusion.
Wannacomet’s Appeal.
The department correctly granted two of Wannacomet’s requests for rulings; viz. that it “is entitled to a fair return on its net investment in property used or useful ... to provide . . . adequate water service”; and that a “fair return is one . . . sufficient to insure confidence in the financial soundness of the utility, adequate to maintain ... its credit, sufficient to enable it to attract the capital required to meet its public obligations, and commensurate with that generally being earned on investments in business undertakings attended by corresponding risks.” See New England Tel. & Tel. Co. v. Department of Pub. Util. 331 Mass. 604, 617; Federal Power Commn. v. Hope Natural Gas Co. 320 U. S. 591, 603. We must examine whether the department’s decision on rate of return and rate base involves confiscation of Wannacomet’s property and whether there is substantial evidence to support the department’s conclusion that the rate base should not include the $76,000 to be transferred from earned surplus to the depreciation reserve.3
[458]*458The Rate Base as Affected by the Depreciation Reserve.
From 1915 through 1921, Wannaeomet, like certain other companies, provided for depreciation by charging income and making annual deductions from the plant account on the asset side of the balance sheet. Depreciation of $26,630 was thus accumulated. By St. 1921, c. 268, § 1 (now G. L. [Ter. Ed.] c. 155, § 5A), the department was given authority, when approving a rate schedule or security or debt issue proposed by a utility to ‘ ‘ order such company to set aside out of earnings such allowances for depreciation and for such period or periods as the department may from time to time prescribe.”
In 1922, Wannaeomet transferred the $26,630 of accumulated depreciation to a reserve account. In 1934, the department directed reversal of this change and it was restated as a deduction from plant account. This, one witness testified, had the effect of understating both the plant account and the depreciation reserve by $26,630. In 1934, also, the department and Wannaeomet agreed that there would be appropriated to the depreciation reserve at least $4,000 annually (see D. P. U. 4751, p. 5, May 22, 1934). A petition of the town’s selectmen for reduction of water rates was denied.
In 1951, the department approved increased rates. Wan-nacomet’s annual depreciation was then ordered increased to not less than $6,000. See D. P. U. 9603, pp. 6, 8. In 1958, a further rate increase was approved and annual depreciation was ordered increased to not less than 1.5%. See D. P. U. 12,254, p. 6. In each case the book depreciation reserve was deducted from plant investment in computing the rate base. In 1958, Wannaeomet borrowed $250,000 upon a twenty-five year note, requiring the department’s approval.4 The record does not disclose whether [459]*459the department approved issuing this note, but to do so it would have had to find that Wannacomet’s surplus plus paid in capital stock and stock premium was then at least $250,000. If surplus had then been charged any major part of $76,000 to make up an alleged deficiency in the depreciation reserve, such a finding could hardly have been made.5 The department’s annual return form (p. 204) required disclosure of the composition of the depreciation reserve and of the charges and credits to it. Returns for the years 1956 to 1961 are in the record.
At the hearings in 1961 and 1962 on the new schedule of rates then proposed, John O. Adams, Jr., a consulting engineer called by Wannacomet, testified concerning the proportions of total revenue that should be derived from general water service and from fire service, respectively. In making certain cost and revenue studies, he had distributed the then depreciation reserve among the various property accounts. For this purpose, he estimated in dollars the total average useful service life of various categories of property and the portion of that service life already incurred. For all plant accounts, he estimated that the total then incurred useful life was approximately $136,000. He applied the then total depreciation reserve of about $60,000 to each of the various plant accounts in the proportion in which his estimate of the then incurred useful life (in dollars) of the property in that account bore to $136,000. The department in its decision in effect adopted the view that this study, devised for allocation purposes, sufficiently established that the depreciation reserve of $60,000 was inadequate by some $76,000; that this amount should be trans[460]*460ferred from surplus to the depreciation reserve; and that Wannaeomet should not he allowed to include this $76,000 in its rate base.
The history of the department’s regulation of Wanna-comet’s depreciation practices outlined above shows some inconsistencies and lack of continuity. Cf. Grant & Norton, Depreciation, 318 (“Where the regulation of prices is a continuous process, as in the regulation of public utility rates, the key to equitable treatment is consistency”). Particularly in the earlier years, this inconsistency may have been caused in some measure by general uncertainty on the part of public utilities and regulatory bodies about (a) the role of depreciation in public utility accounting, and (b) the methods of making effective modern measures of depreciation. Over the years there has been much discussion in regulatory circles of these problems and of how far it is fair or constitutionally permissible to require retroactive adjustments in depreciation reserves, at the expense of surplus, in order to bring the actual reserves into line with' a theoretically more correct estimate of what the reserve should be.6 A dilemma confronts both a regulatory body and a regulated utility, when it is determined that the latter has failed to establish a depreciation reserve equal to that which would have existed if the company from its inception had consistently reserved on a straight line basis. The question then arises whether it is fairer to the public and to the utility to bring the actual reserve more closely into line with the theoretically correct reserve (a) through [461]*461an immediate adjustment at the expense of surplus, or (b) by gradual amortization of the deficiency through increased future depreciation deductions from earnings. It also must be decided, in a rate case, whether, in determining the rate base, the actual depreciation reserve or the larger theoretical reserve should be deducted from the plant account. Professor James C. Bonbright, Valuation of Property, 1140, expresses the view that “the proper answer . . . depends on the circumstances which led to the failure to provide adequately for depreciation. If the company was negligent or if it failed to abide by well-accepted rules of good accounting as those rules were understood at the time, or if its niggardly depreciation policy was in derogation from the [commission’s] accounting orders . . . it should not now be permitted to make up for its bad practice at the expense of future ratepayers. . . . [Then,] the capital outlay that ought to have been amortized should be deducted from original cost. But if the reserve proves to be inadequate through no fault of the company management, or if it is inadequate only when judged by tests of good accounting that were not generally accepted and not prescribed by the commission until a later date, then a fair application of the ‘original-cost’ principle precludes the deduction of depreciation in excess of the reserve actually accrued.” See Bonbright, Principles of Public Utility Bates, 212-216, and Professor Baker’s comments, already cited, in 58 Harv. L. Bev. 886, 893-894.
A view, consistent with that just quoted, is expressed in some decisions7 which consider as important the circumstance that the utility’s book depreciation reserve was developed with the approval of the regulatory commission so that it would be inappropriate to adjust the depreciation [462]*462reserve retroactively by a charge against surplus (and against capital if necessary). Such an adjustment, of course, might deprive the utility of any opportunity to recoup in revenues the amount of the adjustment, which may not have been adequately reflected in rates theretofore charged. In at least one State, however, the regulatory commission has been permitted to require deduction from the rate base of the excess of accrued straight line depreciation over the book reserve. The New York cases so holding may rest in part upon the circumstance that the New York regulatory commission has not had the same direct statutory power, possessed by the Massachusetts department (see G. L. [Ter. Ed.] c. 155, § 5A) to regulate book reserves and depreciation allowances. See Consolidated Edison Co. v. Maltbie, 300 N. Y. 196, 204-206, and cases cited. See also Lippitt, Net Investment Rate Making — The Deduction for Depreciation, 62 Harv. L. Rev. 1155, 1169-1170.
The authorities discussing retroactive reserve adjustments, however, do not suggest hard and fast rules, but at most state criteria for deciding, in any particular instance, what adjustments are reasonable, equitable, and in the long term public interest. See N. A. R. U. C. Report of Committee on Depreciation, 1944. The application of these criteria to the facts of a particular case is, of course, primarily for the decision of the regulatory body. Judicial correction of the action, however, would be appropriate, if the department were to disregard these and other relevant criteria unreasonably, arbitrarily, or capriciously, or to take action relating to the reserve not based upon substantial evidence or resulting in confiscation. See G. L. c. 30A, §14(8).
The department did not indicate why it rejected less severe methods of gradual adjustment of the reserve over the remaining useful life of the properties,8 beyond saying [463]*463that “present customers should not he compelled to pay . . . a return on plant . . . not available to serve them.” It was not suggested that Wannacomet had failed to abide by currently accepted accounting rules or the department’s accounting orders. The department, after summarizing Adams’s testimony, merely stated the conclusion that the reserve “deficiency ... is so substantial on the basis of . . . [Wannacomet’s] own evidence that it can no longer be overlooked.”
In determining whether the department’s action is confiscatory or whether it is not based upon substantial evidence, the manner in which question about the reserve arose is of significance. Early in the 1961 hearings, Wan-nacomet presented sufficient evidence of the amount of its investment in plant and its depreciation reserve as shown on its books. These were shown to have been kept in compliance with the department’s accounting procedures and orders. In the course of the cross-examination of Adams and of another witness in 1961, there took place the discussion, already summarized, about the incurred useful life of the various categories of plant. On the basis of Adams’s general testimony about his study, undertaken for other purposes, counsel for the town then took the position that a “large part” of surplus was not a prudent investment on which a return should be allowed. Wannacomet offered no further explanation of its depreciation reserve and presented no evidence to clarify or contradict Adams’s testimony (except for a brief discussion of Wannacomet’s pre-1922 depreciation practices).
Wannacomet had the burden of establishing its rate base. Adams’s general testimony warranted the conclusion that the book depreciation reserve was, up to a maximum of $76,000, lower than what a straight line reserve would have been. We need not now decide whether a more complete depreciation study (cf. Pennsylvania Pub. Util. Commn. v. Pennsylvania Power & Light Co. 14 P. U. R. 3d 438, 455-457) would have been appropriate before the department affirmatively ordered the reserve increased. For the pur[464]*464poses of this rate case, the department could conclude from Adams’s testimony (at least in the 'absence of explanatory testimony) that the rate base had not been established by Wannacomet to the extent of the then indicated reserve deficiency. We think (see fn. 3, supra) that the record is insufficient to show that the department’s failure to accept a higher rate base will result in confiscation or that it had insubstantial evidential basis.
Upon the evidence, it is not possible to determine whether the department has failed to apply the criteria, already summarized, which should govern retroactive refusals to recognize the adequacy of past depreciation deductions. The department did rule that there was “no substantial evidence which would warrant a finding that in any year prior to . . . [1961, Wannacomet] received a return on its rate base in excess of a fair return. ’ ’ There was, however, little evidence of Wannacomet’s past earnings, beyond the department’s decisions denying a rate decrease in 1934, and granting rate increases in 1951, 1958, and 1961 (interim), and its annual reports for 1956-1961. Wannacomet itself had caused the exclusion from evidence of its annual reports to the department from 1915 to 1955, inclusive. Accordingly, the department was not and this court is not in a position to determine whether Wannacomet had already recouped the indicated reserve deficiency from customers (by rates higher than those which would then have been fair), or whether Wannacomet’s past rate of return has been no more than was fair, with the consequence that the untaken depreciation (causing the reserve adjustment) has not been reflected in determining past rates.
The department’s decision mentions “that when dealing with . . . composite rates of depreciation, there will be times when property having useful lives shorter than the average will have been retired,” thus creating an “apparent deficiency in the reserve, which would be [later] offset . . . by the fact that other property has a” longer than average useful life. This characteristic of composite depreciation rates is sometimes called “infant mortality.” [465]*465See the Potomac Elec. Power Co. case, 55 P. U. R (N. S.) 65, 84, and the Washington Gas Light Co. case, 78 P. U. R (N. S.) 82, 88. Wannaeomet has not presented evidence showing that its plant account and depreciation reserve have been materially affected by such infant mortality.
It may be that Adams, in his computation showing $136,000 of then incurred plant life, took into account the pre-1922 depreciation of $26,630, which had been removed from the depreciation reserve by the department’s 1934 order. Wannaeomet, in effect, so argues. To the extent that Adams did include pre-1922 depreciation in the $136,000, it would seem that the maximum reserve deficiency suggested by his testimony may have been less than $76,000. Wanna-comet did not prove and the record does not show what Adams in fact did in this respect. If the department has overstated the amount of the deficiency, it is within its discretion to reopen the record for clarification and correction of this point.
The Rate oe Return".
Wannaeomet disputes the department’s conclusion that “a rate [of return] in excess of 8 per cent on the equity portion of the rate base” (or a return of 6.5% on the whole rate base) will provide Wannaeomet “a fair return on the equity portion of its capital.”
The department’s method of computation of the rate base and of the rate of return is stated somewhat .indefinitely in its decision. The department appears to have assumed that Wannaeomet, after impending capital adjustments, would have outstanding long term debt of $360,000 (old note for $250,000 and new debt, to be issued, of $110,000) and that the “weighted average cost of long-term debt” would be 4.95%. It finds a rate base of $695,153 (after the $76,000 increase in the depreciation reserve and other minor adjustments). The department proposes to allow Wannaeomet to earn a net operating income of $45,185. It would appear that this would result in a return of between 8.7% and 8.92% upon the equity portion (except [466]*466earned surplus) of the rate base allowed by the department.9 Upon earned surplus, the department could reasonably allow only the composite rate of return on debt and equity (other than earned surplus). See New England Tel. & Tel. Co. v. Department of Pub. Util. 327 Mass. 81, 91, 96-97.
Although much of the earned surplus may have been earned principally through the use of equity capital prior to 1958 (when the $250,000 note was issued), we think that this does not make the New England Tel. & Tel Co. case inapplicable. If the earned surplus was accumulated by the use of an unnecessarily large amount of equity capital at times when Wannacomet could have obtained cheaper debt capital (without having its capital structure depart from a reasonable debt ratio), the department is not required to allow on that surplus more than the composite rate of return that would have been allowed if Wannacomet had in fact employed a reasonable debt ratio while accumulating the surplus. See New England Tel. & Tel. Co. v. Department of Pub. Util. 331 Mass. 604, 611-612, 617-619.
[467]*467Two expert witnesses, called by Wannacomet, testified about the return upon common stock necessary to attract investors in the light of available returns on investments involving comparable risks. The department and the town offered no evidence on rate of return.
John Day, vice-president of The First Boston Corporation, used the system he “would use if . . . negotiating” a price at which he thought he could sell Wannacomet’s common stock to the public. “Normally . . . earnings ratios [to market] of comparable companies are used as a base point, but since there is no quoted market for Wannacom-et’s common stock . . . earnings-book value ratios serve the purpose better. ’ ’ Accordingly, he took published material about this ratio in other companies, and “then considered the special circumstances10 of Wannacomet . . . and made adjustments that an investor would make in determining the price he would pay.”
Exhibits showed that nine selected water companies11 on the average earned 9.37 % on their common book equity for five years, and 8.49% on the market price of their stock, over ten years. Day concluded that an 11% return (disregarding underwriting and financing expense) would be required to attract capital to Wannacomet because an “investor who can obtain earnings that would average 9.5 per [468]*468cent on the equity investment in a group of large, safe water companies with marketable stock, or 8.5 per cent on the price paid for such stock . . . would not consider an [equity] investment in . . . Wannaeomet . . . unless . . . adequately compensated for the significant additional risk. ’ ’
Reis, vice-president of a firm of financial analysts, testified that in his judgment equity investors in Wannaeomet would demand “no less than 10 per cent” upon the “funds it . . . will receive from them.” He supported this conclusion by various studies.12
No evidence suggests to us that Wannacomet stock would obtain any better general market reception than the stocks of the larger, better known water companies used for comparison. Both Day and Reis based their testimony on the return necessary to float Wannacomet’s stock on the general market, notwithstanding the circumstance that its parent company would probably purchase the shares. See New England Tel. & Tel. Co. v. Department of Pub. Util. 327 Mass. 81, 94-95; Salisbury Water Supply Co. v. Department of Pub. Util. 344 Mass. 716, 722-723. The parent company, of course, is not bound to take any new stock and is [469]*469as much entitled as any other potential investor in Wanna-comet to have that company earn a fair return.
The department stated that its proposed “rate ... is calculated to allow ... in excess of 8 per cent on the equity portion of the rate base. This return on common is equal to or greater than the returns consistently earned by many . . . companies . . . selected for comparison . . . and substantially improves on the return . . . earned ... in recent years during which, despite a lower return . . . [Wan-nacomet] succeeded in issuing new capital stock.” We assume that this last reference is to new stock (which with the premium amounted to $25,025) issued in 1958 to the parent company, presumably to effect a 50% debt ratio (see fn. 5, supra) at the time the now outstanding $250,000 note was issued. The language, “returns consistently earned by many . . . companies . . . selected for comparison,” may refer to three of the nine companies used for comparison by Day (fn. 11, supra), unless reference is also intended to Reis’s tabulation of the ratio of earnings to market price disregarding financing costs (fn. 12, supra).
It is difficult to determine by comparison with larger companies the fair return on equity investment for a small water company like Wannacomet whose stock is wholly held by a parent company. Wannacomet in various respects may differ from the selected larger companies, some of which are in other parts of the nation. Nevertheless, even if general market comparisons necessarily are in a degree unsatisfactory, they may be the most helpful available indicia of market conditions. Indeed, it would probably be difficult or impossible to produce, for any substantial group of small water companies closely comparable to Wannacomet, exhibits showing the relation of net earnings available for equity to the proceeds from stock purchased by parent companies and institutional investors, and the extent to which financing charges, normally borne by the utility itself, may be transferred to a parent company or be reduced in the case of a stock sale to an institution. Giving full weight, however, to the reliability of the basic pub-[470]*470listed investment data used by Day and Eeis, the department was not bound to accept their opinions and conclusions based upon such data. It could appropriately consider the likelihood, mentioned in the evidence, that any new Wanna-comet stock would be issued to the parent company or to a few institutional investors at low financing costs, and the possibility that such purchasers might be more likely than the general public to pay a price near to the intrinsic value of the shares, because fully informed about the company, and thus less apt to be influenced by adverse factors of the type mentioned by Day (fn. 10).
The department should have made more complete subsidiary findings. See Herson’s Case, 341 Mass. 402, 407-408; Bay State Harness Horse Racing & Breeding Assn. Inc. v. State Racing Commn. 342 Mass. 694, 701; Massachusetts Medical Serv. v. Commissioner of Ins. 344 Mass. 335, 339-340; S. C., ante, 346, 347; Leen v. Assessors of Boston, 345 Mass. 494, 502; Packard Mills, Inc. v. State Tax Commn. 345 Mass. 718, 723-726; Hamilton v. Department of Pub. Util., ante, 130,137. Such findings are particularly important and helpful where, as here, there is a long record (about 800 pages of transcript and 50 exhibits). If the department relied at all upon its own staff, their computations and opinions should have been introduced in evidence, and, if judicial notice was taken of any facts, there should have been compliance with G. L. c. 30A, § 11 (5). See Salisbury Water Supply Co. case, 344 Mass. 716, 721-722.
Because I believe that the department’s decision on this phase of the case rests on too meager subsidiary findings, I (before passing at all upon whether the rate of return is confiscatory) would remand the case to the department for more specific findings (see G. L. c. 30A, § 14 [8] [e] and [f]) on rate of return, and, if necessary, for the taking of additional evidence, as in the Salisbury Water Supply Co. case, 344 Mass. 716, 720-723. A majority of the court, however, are of the opinion that the case may be disposed of without such a remand. They think that the evidence relied upon in Wannacomet’s brief shows at most a small dis[471]*471crepancy (within the range from a fraction of one per cent to two percentage points) between (a) the return on equity allowed by the department and (b) that indicated by comparative statistics. This discrepancy, they feel, .has not been shown to be so large, or so clearly to exist, as to establish confiscation.
The majority give weight, among other things, to (a) the various elements of dissimilarity between Wannacomet and the larger companies compared with it; (b) the remoteness from New England of some of these companies; and (c) the probability that, for parent company or institutional purchases of stock in the amounts likely to be issued by Wan-nacomet, financing charges may not be as large as estimated by Reis (see fn. 12), thus rendering more significant his statistics (unaffected by financing charges) of the ratio of earnings per share to market price per share for other water company stocks. In reaching their conclusion, the majority judges point out that it will be open to Wanna-comet to file new rate schedules if the allowed return turns out to be inadequate after further experience with meters, with the rates now to be allowed, and with efforts at new financing.
A final decree is to be entered affirming the decision of the department.
So ordered.