In January, 1879, Mrs. Hannah P. Dimick made and executed what purported to be her last will and testament. It was first signed by her in the presence of only one witness, and subsequently taken to the second witness, and after that to the third, each of whom signed as a witness in the absence of the other two. This paper remained without change until the death of the testatrix on the sixth day of February, 1888, and afterwards on the third day of March, 1888, the same was presented to the probate court and there approved as her last will and testament. Upon appeal to the Superior Court the decree of the probate court was reversed, and the appellee now brings the case to this court for the revision of alleged errors in the Superior Court.
The sole question is—whether the instrument was invalid-as a will for want of proper attestation. The answer depends upon the construction and effect of our statutes relating to the execution and attestation of wills.
The difficulty however is not so much in determining the meaning of our statutes, as in ascertaining which of two different statutes applies. If we take the statute in force when the instrument was executed, which positively required the witnesses to sign in the presence of each other and of the testator, the attestation was clearly contrary to law and the will was void. If on the other hand we take and apply the statute as it stood at the decease of the testatrix, which was first passed in 1885, the attestation was legal and the instrument valid as a will.
The statute in force when the will was attested provided that “no will or codicil shall be valid to pass any estate, unless it be in writing, subscribed by the testator, and attested by three witnesses, all of them subscribing in his presence and in the presence of each other; but no will of personal estate made before the twenty-seventh day of June, 1848, shall be invalid if not so attested; and all wills executed according to the laws of the state or country where they were executed may be admitted to probate in this state.” General Statutes of 1875, p. 369, sec. 2.
*186The statute in force at the death of the testatrix provided that “ no will or codicil shall be valid to pass any estate, unless it be in writing, subscribed by the testator, and attested by three witnesses, each of them subscribing in his presence; and all wills executed according to the laws of the state or country where they are executed, may be admitted to probate in this state, and shall be effectual to pass any estate of the testator situated in this state.” General Statutes of 1888, § 538.
As a will is ambulatory during the life-time of the one executing it, and no rights can vest under it till the death of the testator, it must be conceded to be within the rightful authority of the legislature, until the death of such testator, by retroactive legislation to change the formalities previously .prescribed for the due execution of wills, and to affect every instrument previously executed, making it valid or invalid as the case may be. But it is one of the fundamental canons of construction accepted everywhere, and most firmly held by this court, that a statute shall always be interpreted so as to operate prospectively and not retrospectively, unless the language is so clear as to preclude all question as to the intention of the legislature. Brewster v. McCall's Devisees, 15 Conn., 274; Goodsell's Appeal from Probate, 55 Conn., 171.
In the light of this well settled rule it would seem impossible to give the last mentioned statute a retroactive operation, for not only is such intent not made clear but the contrary more clearly appears. All the provisions look forward rather than backward. But suppose the act is not clear either way, are there any other principles which may control and enable us to determine whether the validity of the execution-of a will should be determined by the law existing at the execution, or as it was at the death of the testator.
Upon this question there is a disagreement among the authorities, to which we may hereafter refer, but at present we will look only at the reasons for the differing opinions. The reasons for applying the later statute, stating them as *187strongly as possible, are as follows:—“As until the death of the testator the paper executed by him, expressing his wishes, is not a will, but a mere inchoate act which may or may not be a will, the law in force at the testator’s death applies and controls the proof of the will.” Sutton v. Chenault, 18 Geo., 1. This in substance, is the reasoning of all the courts that have accepted the doctrine that the validity of a will must be determined by the law existing at the death of the testator.
But as plausible as the reasoning may seem we think it is fallacious, at least as applicable to our'laws. The act of bequeathing or devising is something more than inchoate or ambulatory. On the other hand it becomes a completed act when the will is executed and attested according to law, although it does not take effect on the property till a future time. A power of revocation is retained for life, but even that, to be effectual, must be exercised in the way and manner prescribed. What need could there be of any revocation, with, prescribed formalities, if the executed paper is to be considered only as an expression of the signer’s wish and is in no sense a will. The latest wish, however expressed, ought to overcome the former.
The theory of our statutes seems to us directly opposed to the reasoning referred to, however it may be with the statutes of other jurisdictions. Certain formalities of execution and attestation are prescribed as prerequisites to the validity of a will, and without compliance with which it is no will at all, although it is clearly a wish. In terms it is declared to be incompetent to pass the title to property at the death of the testator. The precise language is:—“ No will or codicil shall be valid to pass any estate unless,” etc. So that our statute amounts to a positive rule for the transmission of property, which must be complied with, as a complete act at the time of execution, or never, so far as the act of the testator is concerned. And what the statute makes positively bad can only be made good by positive legislation, manifestly retrospective.
Our law in effect says to every person who would make a *188valid disposition of his property by will, that he must observe the specified formalities, and if he complies, it contains an implied assurance that he shall not be disappointed and defeated in his purposes by any subsequent change, unless the new law clearly affects wills previously made.
That such is the true theory of our statutes is rendered still more manifest, we think, by reference to the provision, which has for more than a century existed, that at the time of executing a will or at any time during his life, the testator may obtain the affidavit of the witnesses of tire facts required to prove it in court, which having been written upon or attached to the will, “ shall be accepted by the court of probate, as if it had been taken before said court.” General Statutes of 1784, p. 264, and of 1888, sec. 545. It has thus been the settled policy of our law in effect to assure the person who executes a will, not only that the prescribed formalities shall be sufficient and controlling, but that even the evidence of the proper execution shall be sufficient finally to establish it, though many years shall have intervened. Under such a system surely it cannot be said that what the testator has thus prepared for his will is only a mere paper upon which he has indicated simply his wishes at the time as to the final disposition of his property.
If we have rightly apprehended the true theory of our statute laws no further vindication of the ruling of the Superior Court is necessary. If however our views find confirmation in the decisions of the courts of other jurisdictions, we may well feel greater assurance that we are right.
This subject has not been much discussed in the courts of the United States, and the few decisions that may be found on the subject are not harmonious, as we have before stated. The text writers make only a brief reference to the question. Of these, Jarman and Redfield favor the doctrine that the law existing at the death of the testator controls, rather than that at the date of execution. 1 Jarman on Wills, p. 337; 1 Redfield on Wills, 4th ed., side p. 408. In 1 Williams on Executors, 4th ed., top page 94, note, the question is left as *189doubtful. In Schouler on Wills, § 2, and in 1 Rev. Swift’s Digest, 140, the contrary view is held.
The courts of New York, South Carolina and Georgia adopt substantially the proposition as laid down by the text writers first mentioned. Lawrence v. Hebbard, 1 Bradford, 252; Houston v. Houston, 3 McCord, 491; Elcock’s Will, 4 McCord, 39; Sutton v. Chenault, 18 Geo., 1; Hargroves v. Redd, 43 Geo., 142.
But the courts of England and of Vermont and Pennsylvania have adopted substantially the principles for which we contend. Downing v. Townsend, Ambler, 280; Gillmore v. Shooter’s Exr., 2 Mod., 310; Giddings v. Turgeon, 58 Verm., 106; Mullen v. McKelvy, 5 Watts, 399; Taylor v. Mitchell, 57 Penn. St., 209.
The difference of opinion referred to may perhaps be attributed in some measure to a difference in the language and theory of the local statutes. For instance, in Lawrence v. Hebbard, supra, the surrogate, while he adopts the line of reasoning referred to as contained in Sutton v. Chenault, supra, yet he emphasizes the fact that the statute of New York in terms refers to wills previously executed and says they shall not be made void, and thereby contains an implication that they may be made good by the latest statute, and so the court holds.
The courts of South Carolina confine their decisions to wills of personal property. Referring to the distinction between wills as to personal and real estate, they intimate that as to the latter the law existing at the date of execution may possibly control, upon the ground that to transfer the title to real estate the existing statutory regulations may possibly control. But our statute practically abolishes any such distinction by putting wills of personal estate upon the same ground as those affecting realty, so far as the formalities of the execution of the will are concerned; the language being “no will shall be valid to pass any estate” unless, etc.
The counsel for the appellee contended that the case of Giddings v. Turgeon, supra, had no application, because the *190question there was so different from the one in the case at bar. While conceding the difference between the two cases we still think the principle adopted by the court is applicable. The validity of the entire will was there under consideration, in which one of the legatees was a married woman, whose husband was one of the witnesses to the execution of the will. As the law of Vermont then stood he was incompetent as a witness, and it was improperly attested. By a subsequent act, however, a husband was made a competent witness, where a legacy had been given to his wife, to prove and establish the will, although the legacy to the wife would not be good. It was, contended that the subsequent law would apply, but the court held that the will was not good, because it must be proved as the law required at the time of its execution, which is the principle contended for here.
In Taylor v. Mitchell, supra, the point made was that the will there in question was void under the statute in force when the testator died. The case was very well considered and in the opinion of the court by Sharswood, J., it is said:—“ Retrospective laws generally, if not universally, work injustice, and ought to be so construed only when the mandate of the legislature is imperative. When a testator makes a will, formally executed according to the requirements of the law existing at the time of its execution, it would unjustly disappoint his lawful right pf disposition to apply to it a rule subsequently enacted, though before his death. While it is true that every one is presumed to know the law, the maxim in fact is inapplicable to such a ease; for he would have an equal right to presume that no new law would affect his past act and rest satisfied in security on that presumption. * * * It is true that every will is ambulatory until the death of the testator, and the disposition made by it does not actually take effect until then. General words apply to the property of which the testator dies possessed, and he retains the power of revocation as long as he lives. The act of bequeathing or devising, however, takes place when the will is executed, though to go into effect at a future time.”
*191No case hitherto has been before this court involving the precise question now under consideration, but analogous questions have been discussed in several cases, and principles applied to them which naturally and logically lead to the result we have reached in this case.
In Brewster v. McCall’s Devisees, 15 Conn., 274, one question was whether the will, executed in 1826, by a person who died in 1838, could convey real estate purchased after the will was executed and before the act of 1831, which provided for the devising of real estate not owned by a testator at the time of executing his will but acquired afterwards. The court, in refusing to apply the act existing at the death of the testator, said:—“ The general rule is that statutes shall not be construed retrospectively, unless by their express terms or otherwise such appears to be the manifest intent of the legislature.. * * * Here there is nothing from which we have a right to infer that it was intended to affect any wills which had been executed prior to the passing of the act.”
In Gaylor’s Appeal from Probate, 43 Conn., 82, the question was whether under the statute existing when the will was made (Gen. Statutes of 1866, p. 402,) witnesses must subscribe in the presence of each other. Carpenter, J., in giving the opinion of the court says, on p. 85:—“ The language of our statute existing when the will was made is explicit and entirely free from ambiguity. It only requires that all the witnesses shall subscribe their names in the presence of the testator. * * * In the revision of 1875 we find, in addition to the statute as it previously existed, the clause inserted, ‘ and in the presence of each other.’ But that cannot affect the validity of any will previously executed.”
In Goodsell’s Appeal from Probate, 55 Conn., 171, the will was made in January, 1871, and the testator married in May following, and died in 1886, leaving a widow, but no child had been born to him. When the will was executed and when the marriage took place the common law was in force which required both a subsequent marriage and the birth of *192a child to revoke a will. In 1885 an act was passed providing that “if after the making of a will the testator shall marry, or if a child is born to the testator and no provision is made in the will for such a contingency, such marriage or birth shall operate as a revocation of such will.” It was held that the later statute did not apply, and that “as a rule of interpretation all statutes are to operate prospectively unless they contain language unequivocally and certainly retrospective.”
For the foregoing reasons we conclude there was no error in the judgment of the Superior Court.
In this opinion Park, C. J., Pardee, and Beardsley, Js., concurred. Carpenter, J., dissented.