Knowlton, J.
This is a petition to the Probate Court for a revocation of a decree allowing the will and codicil of Alford B. Wilton, and for an allowance of them with certain portions of the codicil stricken out, as having been improperly inserted after the paper was executed. The' material portions of the codicil are as follows : “ Whereas by my last will and testament dated the eleventh day of April, 1888, in article tenth I gave to my sons and grandchildren the income of rest and residue of my property, after paying my debts and legacies, to be equally divided between them. I do hereby revoke the said legacy, and instead thereof I give and bequeath the income of the said rest and residue of my property as follows : One eighth of the same to my son John George S. Wilton, the other seven eighths to my son Augustus W. Wilton, and my grandchildren, to be divided equally between them. I also give and bequeath to Cornelia E. Lawrence the sum of five hundred dollars.” The words “ income of ” in the first sentence, and the words “ income of the ” in the second sentence, are written in the margin before the words “rest” and “said,” respectively, each of which latter words is" at the beginning of a line. These words in the margin and the words “E. Lawrence” in the second clause are in the handwriting of one Knapp, the scrivener who wrote the codicil and who had previously written the will, and are in a different ink from the rest of the codicil and written by a different pen. Knapp testified that" the words in question were written before *255the execution of the codicil, and as tending to explain the use of a different pen and ink, he testified that he had owned and had used considerably a fountain pen; but no one testified to seeing it at the time of the execution of the codicil, and there was no testimony that there was more than one pen or one bottle of ink in the room at the time. The codicil was written in 1888, and it was evident from the testimony of Knapp, and from his statements at different times, that he had no particular recollection in regard to the making of the first two changes in the codicil, although he testified that when writing the bequest to Cornelia the .testator did not remember her married name, and that a little space was left, in which the words “ E. Lawrence ” were after-wards written when the testator ascertained the name from his son Augustus before the instrument was executed. The codicil correctly describes the bequest in the original will. The appellant offered to prove that, about two weeks before the codicil was made, the testator stated to him how he had left his property by his will, and recited the tenth clause of the will exactly as he thought it was, and that he stated it as it would have been described in the codicil if the words in question had been omitted. The appellant’s first exception is to the exclusion of this testimony. The offer was in substance to show, by a previous declaration of the testator in regard to the disposition of his property by a will then existing, that he understood the will in a way which would make it improbable that he would refer to it as he did in the codicil as it was finally written. For some purposes previous declarations of a testator indicating a state of mind or an intention in regard to the disposition of his property are competent when the validity of a will is in question. Subsequent declarations, for the purpose of proving facts bearing upon the validity of a will, are not ordinarily received. Doe v. Palmer, 16 Q. B. 747. In re Adamson, L. R. 3 P. & D. 253. Smith v. Fenner, 1 Gall. 170. In Sugden v. St. Leonards, L. R. 1 P. D. 154, it was held, after much consideration, that subsequent declarations of a testator may be received to prove the contents of a lost will. If we assume in favor of the appellant, without deciding, that a testator, making declarations to a relative in regard to the disposition of his property by an existing will, will probably state the truth as he understands it, so that the *256statement propierly may be received as evidence of bis belief, the question as affecting the validity of this codicil is not what the testator thought at the time of making the declaration, but what he thought at the time of executing the codicil. If such a declaration is offered, the question immediately arises whether in length of time or in the occurrence of events it is so closely connected with the making of the codicil as fairly to indicate the state of mind of the testator when the codicil is executed, or whether from lapse of time or from the occurrence of intervening events it has ceased to have probative force as to the condition of the testator’s mind when he signed the codicil. In the present case we think the judge, in the exercise of his discretion, might well hold, in view of other facts, that the evidence was too remote. It appeared that, because one of his sons came to his room in a condition showing that he had been drinking intoxicating liquors, the testator determined to change the tenth clause of his will. He accordingly sent for Knapp, the scrivener, who had the will in his charge, and requested him to bring it. Knapp came with the will, took it to the testator’s room, was told by the testator that he wished to change his will, and there prepared the codicil. The only inference from these facts is that the will was brought by direction of the testator that it might be examined and used in making the codicil. The will was there in the possession of the testator and of the scrivener during the preparation and execution of the codicil. The codicil refers to the clause to be changed in the will by its number, and in this way indicates that the will was then examined. The judge might well find that, if there was any previous forgetfulness or mistake of the testator in regard to the part of the will affected by the codicil, there was no probability that it continued through the interview with Knapp and up to the time of the execution of the codicil, and he well might exclude testimony tending to show that the testator was mistaken about it two weeks before. This exception must be overruled.
The appellant called Knapp as a witness, and then offered evidence of statements made by him tending to contradict his testimony, without previously calling his attention to the statements. This evidence was rightly excluded. Pub. Sts. c. 169, § 22. Ryerson v. Alington, 102 Mass. 526. Batchelder v. *257Batchelder, 139 Mass. 1. The appellant contends that because Knapp was a witness to the will and. codicil the statute should not apply to him. In this case he was not called as a witness to prove the will or codicil. The will and codicil were proved and allowed many years ago, and he was called by the appellant to show that the decree of the Probate Court was erroneous. There is no reason why the ordinary rule should not apply to his testimony.
The plaintiff asked the court to rule, “ that the words in the margin being unnoted, and being in different ink, the presumption of law in this case is that they were added after execution.” There is no presumption of law in a case of this kind as to the time when the disputed words were written. The question is one of fact to be determined on all the evidence. Ely v. Ely, 6 Gray, 439. The burden of proof is on a party presenting a will to show that alterations or interlineations in it were made before its execution, and unless there is something in the nature of them, as applied to other parts of the writing, or unless there is other evidence to show that they were a part of the will when it was executed, they will be rejected for want of proof. The opportunities for making alterations in a will are often so great, and the requirement of the statute that a will shall not take effect unless it is executed with due formality is so strict, that apparent alterations in a will are looked upon with some suspicion. The validity of changed portions is not assumed, but there may be evidence to establish it; and this evidence may be found in the instrument itself, or may come from outside of it. Greville v. Tylee, 7 Moore P. C. 320. Doe v. Palmer, 16 Q. B. 747. In re Adamson, L. R 3 P. & D. 253. Williams v. Ashton, 1 Johns. & H. 115. In re Cadge, L. R 1 P. & D. 543. Birch v. Birch, 1 Rob. 675. In re Swindin, 2 Rob. 192. In re Birt, L. R. 2 P. & D. 214. In re Penniman’s Will, 20 Minn. 245.
In the case at bar, the codicil with the disputed xvords in it correctly describes the bequest in the original will, and conforms to the will. If the words were omitted, a comparison of the codicil with the will would show a mistake apparent on the face of the paper. This fact tends to control the ordinary inference against a changed portion of a codicil. Anything in the paper itself, as well as in the evidence from outside, which makes it *258more probable that the changes were made before the execution than that they were made after execution, is sufficient to give the alterations validity. In re Birt, In re Cadge, Birch v. Birch, and In re Swind in, ubi supra.
There is another consideration which justifies the refusal to give the ruling. The request was to rule that there was a presumption of law in this case. Now this ■ case is not one in which a codicil is presented for probate. It is one in which a decree, of a court allowing a will and codicil remained unchallenged for more than ten years before this petition was filed. The petitioner assumes the burden of showing that the decree is erroneous. All presumptions are in favor of the decree, and it cannot be set aside without clear proof. It is plain that “ in this case ” there is no presumption of law against the validity of the disputed words in the codicil.
jExceptions overruled,.