Dissenting Opinion by
Mr. Justice Allen M. Stearne:I am unable to join in the majority opinion. The single question involved is whether or not a notation on an unexecuted, altered, carbon copy of a probated will constitutes a revocation of the will under the provisions of the statute. The majority hold that it does. I am of the contrary view.
At the inception it is to be borne in mind that the privilege of disposing of an estate by will or through inheritance is not a natural or a constitutional right but exists solely by grace of the sovereign: Strode v. Commonwealth, 52 Pa. 181; Maginn’s Estate, 278 Pa. 89, 98, 122 A. 264; U. S. v. Perkins, 163 U. S. 625, 16 S. Ct. 1073. It is the sovereign which prescribes by statute the manner in which such disposition shall be accomplished. Even though testator’s intent to make a certain testamentary disposition clearly appears, such intent is totally ineffective unless embodied in a will made and executed in accordance with the terms of the statute: Maginn’s Estate, 278 Pa. 89, 91, 122 A. 264; Baldwin Will, 357 Pa. 432, 440, 55 A. 2d 263. As Justice (later Chief Justice) .Kephart said in Maginn’s Estate, supra, p. 91: “It is possible, in some cases, a ‘decedent may have thought he had made a will, but the statute says he had not. The question is not one of his thought in that respect, but what he actually did, or . . . failed to do ... . “It may happen [that] .• . . wills truly, expressing the intentions of the testators are made without observations of the required forms; whenever, that happens, while the genuine intention is frustrated . . .• the legislature . . . has thought it best-, *482and has therefore determined, to run the risk of frustrating that intention ... , in preference to the risk of giving effect to or facilitating the formation of spurious wills, by the absence of forms. . . The evil of defeating the intention ... is less than the evil probably to arise by giving validity to wills made without any form. ..” ’, or in derogation of testator’s wishes, fraudulently imposing spurious wills on his estate: Churchill’s Est., 280 Pa. 94, 101.” For other examples of situations in which the expressed, intent of a decedent is frustrated because it runs contrary to a positive rule of law see: Lockhart’s Estate, 306 Pa. 394, 401, 159 A. 874; Hartman’s Estate (No. 1), 320 Pa. 321, 330, 182 A. 234. For this reason it has long been the rule that the intention of the testator may not be determined from parol testimony. Such intention must be ascertained from the words of the testamentary writing, executed in compliance with the statute: Iddings v. Iddings, 7 S. & R. 111; Willard’s Estate, 68 Pa. 327. Justice Kephart expressed it thus in Reinheimer’s Estate, 265 Pa. 185, 189, 108 A. 412: “The intention of the testator must be found from what appears upon the face of the will, and, while extrinsic evidence may be admitted to aid or explain, it must always relate to that which is embodied in the will. It cannot have the effect of remodeling the will. The controlling principle regarding the admission of such testimony is that it cannot be received as evidence of testator’s intention Outside of and independent of the written words employed. The court must find its meaning, if there is one, and not, under guise of a construction or under general powers of equity, to assume to correct or redraft' the will, in which the testator has expressed his intentions.” (Emphasis supplied)
'••• This philosophy .applies equally to revocation. Valid wills may only be set aside in the manner prescribed *483by statute. In the early case of Clingan v. Mitcheltree, 31 Pa. 25, we said (p. 36): . . that the statute should have a strict construction, is clearly established by our own cases of Dunlop v. Dunlop, 10 Watts 153, Cavett’s Appeal, 8 W. & S. 26, Greenough v. Greenough, 1 Jones 496, and Lewis v. Lewis, 2 W. & S. 455. . . . There is certainly great safety in adhering to the words of the statute; besides which, is there not a want of power to interpolate other words, or to find equivalents for the express direction of the statute? At all events, there is danger in establishing exceptions to a statutory rule, which, like the present, has been found to be essentially necessary, for the safe enjoyment and secure transmission of real estate, by the experience of ages; for if exceptions once begin, no one can say when, and where, they will end.” And in Williams’ Estate, 336 Pa. 235, 9 A. 2d 377, Justice (later Chief Justice) Schaffer said for the Court (p. 237) : “Under similar provisions in earlier acts it was held that unsigned writings, on the margins, back or other parts of the paper on which the will was written, even though written after the will was executed and expressing an intention that the will be revoked, were not effective, and that the will should be probated and the subsequent writings disregarded: [citing cases].”
This rule is one of general application. In re Aker’s Will, 74 App. Div. 461, 77 N.Y.S. 643, aff’d 173 N.Y. 620, 66 N.E. 1103, involved an attempt to revoke by a signed writing on the margin of the original will. The court held it ineffective because not witnessed as required by statute saying (p. 647): “Where a will is sought to be revoked solely, by writing, it must con: form in that respect to the requirement of the statute; and, failing in that, it does not revoke the will, even though there may be a clear intention so to do. In the present' case the' testator attempted to-revoke his *484will by written words. His act was clearly not in compliance with the statutory requirements, and therefore it failed of effect.” See also: 1 Page on Wills, (3d ed.) Sec. 457 et seq.
As stated in the majority opinion, Susanna Hoover Kehr executed a will dated August 7, 1942, which had been prepared by her attorney. She left the original will with him for safekeeping and retained for her own purposes the unexecuted carbon copy. By the terms of her will testatrix gave the larger portion of her estate to charities. She made no provision for her daughter.
Testatrix died on July 2, 1951. Shortly thereafter her will dated August 7, 1942, was duly admitted to probate and letters testamentary were granted to the executor named therein.
Subsequently, the unexecuted carbon copy of the will which the decedent had retained was produced by her daughter. At the top of the first page, above the typing and written by testatrix in ink, appears the notation “Null and void” and immediately under these words, the letters “S. H. K.” (her initials). Paragraphs three, five, and ten of the carbon copy are crossed out, and on page one of the carbon copy are two corrections in her handwriting. Contending that the will was revoked by the notation which appears on the unexecuted carbon copy, decedent’s daughter took an appeal from probate.
The single • question involved, as above stated, is whether or not this notation appearing on the unexecuted carbon copy of decedent’s will is another-writing sufficient- in and..of--itself .to- revoke the- will which is otherwise yalid...-.- ......-.- •
■Section 5 of the Wills Act of .1947 (April. 2.4, 1947, P. L. 89, 20, PS. 180.5) ..prescribes three-methods:--by which .a will or-.codicil in writing may be revoked;-viz., *485(1) by some other will or codicil in writing; (2) by some other writing declaring the same, executed and proved in the manner required of wills; and (3) by an act to the document, i.e., by being burnt, torn, canceled, obliterated, or destroyed with the intent and for the purpose of revocation.
In considering whether an attempted revocation meets the requirements of the Wills Act, supra, a sharp distinction must be drawn between the three statutory permissible methods of revocation. (1) Revocation may be by terms of another will. (2) To effect revocation by “some other writing,” the words of that writing must expressly declare that the will is revoked. (3) Cancellation is accomplished by an “act to the document [will]” itself.
I have carefully considered the legal effect of the notation which was found on the unexecuted carbon copy which the majority term a “declaration of revocation.” The writing consists merely of three words: “null and void” — two adjectives and a conjunction. It omits any verb. If these words, and nothing more, had been written on a blank piece of paper no one could seriously contend that parol evidence should be received to establish that decedent had thereby intended to say, “I hereby declare that my last will and testament is null and void.” This would be clearly equivalent to revocation by parol. Every legal authority would unhesitatingly condemn it, and there could be no justification for introducing such evidence under the guise of explaining an ambiguity. There would be no words to interpret. The document would fail to operate as a revocation irrespective of decedent’s true intention to the contrary: .....
I freely concede that if the same words had been written on the face- of a valid will, such, will would be effectively revoked. This would be no declaration Of *486revocation nor would the words be ambiguous. It would clearly constitute a cancellation. Such notation would fall in the third category of revocations prescribed by the statute. Such cancellation unquestionably would terminate the legal efficacy of the instrument on which it was written.
With the above stated hypothetical cases in mind, considering only the question whether or not the notation involved in this case meets the statutory standard (.without regard for any expressions of revocatory intent outside of the writing itself), the character of the notation is readily apparent. In and of itself it declares nothing, and therefore cannot be “some other writing declaring [revocation].” It is an act of cancellation. It cancels the document on which it was written. If this unexecuted document never possessed legal efficacy, I agree that testatrix accomplished nothing by her notation, and therefore did a useless thing. But irrespective of testatrix’s real intention this does not justify the admission of parol testimony to prove that she meant to cancel her original will, any more than it would be proper to show that the writer of an unsigned will really meant it to be his will but neglected to execute it.
The majority have quoted from Gray Will, 365 Pa. 411, 76 A. 2d 169, as follows: “. . . the declaration of revocation need not be express, it may ... be by necessary implication.” They have omitted the succeeding explanation of that remark in Gray -Will: “An earlier will- can be revoked by -a later' will, or other writing which-expressly revoke's it,-Or:which disposes..:of an estate: in ::an- .entirely different ¿manner: than the earlier will and is inconsistent therewith. There is no. difference in'principle:'between a' revocation', by ah 'express clause or .by"inconsistent provisions: [citing', cases].'” ^Emphasis- supplied ) Revocation, by inconsistent..proyi*487sions lias long been sanctioned, but no other instance of revocation by implication has ever been upheld. There is a tremendous difference in principle between implying revocation from inconsistency and implying revocation of an original will because of cancellation of an unexecuted and altered carbon copy. Furthermore, this discussion in Gray Will, supra, was clearly dictum, since we only decided that the alleged revocation by a later will had not been properly executed and therefore we had no occasion to consider the sufficency of the language employed.
The majority cite Ford’s Estate, 301 Pa. 183, 151 A. 789, and Burtt Will, 353 Pa. 217, 44 A. 2d 670, where parol testimony was admitted. But in both of these cases the controlling issue was revival (not revocation) of wills. In the Ford Estate the will containing an express clause of revocation was revoked by tearing or destroying the signature of decedent. Testimony was offered to prove the fact that cancellation was done at testator’s direction and also to prove testator’s intention to revive the earlier will. In Burtt Will the revocation was accomplished, as in Gray Will, supra, by inconsistent provisions in a later will. In both cases revocation toas conceded and the question of revival Avas the sole issue. In the present case the question at issue is one of revocation only; viz., Was a revocation established under the permissive terms of the statute?
The majority obviously fail to distinguish cases involving words which have an equivocal meaning and those cases Avliere the intention of testator cannot be determined because he has omitted language to express his thoughts. The alleged revocatory notation made no reference Avhatsoever to the original will. Nor is there any attempt to incorporate it by reference. If language is ambiguous, extrinsic evidence may be ad*488mitted to explain tlie ambiguity, but it must always relate to that which is embodied in the writing. The controlling principle regarding the admission of such testimony is that it cannot be received as evidence of testator’s intention outside and independent of the written tvords employed: Reinheimer’s Estate, 265 Pa. 185, 108 A. 412.
Extrinsic evidence may not supply language which testator has omitted. If the testator’s intention cannot be ascertained because of the omission of language, the alleged revocatory writing must fail of its purpose. The inquiry is confined to the meaning of what the testator has said, and does not extend to the consideration of what he might have said but did not: Nebinger’s Estate, 185 Pa. 399, 39 A. 1049; Conner’s Estate, 346 Pa. 271, 29 A. 2d 514. Interpretation is never permitted to assume the proportions of reformation: DeSilver’s Estate, 142 Pa. 74, 21 A. 882; Jacob’s Estate, 343 Pa. 387, 22 A. 2d 744. It is true in construing ambiguous language of a will we must put ourselves in testator’s “arm chair” in order to determine his intention but we are still required to declare the meaning of the language used, and not some other possible but undisclosed purpose: Yate’s Estate, 281 Pa. 178, 126 A. 254. If intent cannot be gathered from the writing itself, words cannot be supplied to disclose it: McKeehan v. Wilson, 53 Pa. 74; Rouse Estate, 369 Pa. 568, 573, 87 A. 2d 281. The only words which Susanna Kehr used were “null and void.” The majority are obviously supplying words for testatrix when they determine from extrinsic evidence that she meant to say but did not: “I hereby declare that my last will and testament of which this is a copy is null and void.”
For these reasons I dissent. I would affirm the decree of the learned court below.