¶1Opinion by
¶2The case turns on the devise contained in the will of Philip Bender made March 13, 1895, and probated November 25, 1896. Testator gave; first, to his wife a life estate in all his property. Then, following a description of the particular premises here in dispute, he directed as follows, “House and lot to go to Johannes Bender or his children.” Johannes was a son of the testator, and the plaintiff in the action below is a son of Johannes, by a first wife. Testator’s widow died in 1904, Johannes surviving. The latter died October 23, 1905. By his last will duly probated he devised the lot in dispute 'to the defendant, who was his second wife. This sufficiently indicates the status of each of the several parties to the controversy. Philip Bender, whose will we are considering, left two daughters, both married, and three sons also married. He devised to each, in one form or another, a specific piece of property. The devise to each daughter was to her individually, unaccompanied by any words indicating that anything short of an absolute estate was intended by the gift. The several devises to the sons were alike in form, but differed *611from the devises to the .daughters in that in each case the devise was to the son, "or his children.” We mention this circumstance only to note that Johannes acquired the same estate in the lands devised to him that his brothers acquired in the land devised to them respectively; and the further fact, that for his own purposes, the testator clearly distinguished between the estates given his daughters and those given his sons. The case presents a single question, What estate or interest did Johannes acquire under the devise to him "or his children”? If a fee simple estate contingent upon his surviving the testator or the life tenant, as the case might be, then it follows that the estate became absolute in him, and passed to the defendant as his devisee. If, on the other hand, he took but a life estate, then the fee passed directly to the plaintiff for his proportional interest therein, whatever that might be. The action was ejectment; and a verdict was directed for the plaintiff, the court reserving as question of law the effect tó be given the devise. A motion for judgment non obstante was subsequently dismissed, the court holding that the word 'or’ as it occurs in the devise should be read 'and’; and that so corrected the devise gave to Johannes simply a life estate. We are not concerned to inquire what estate Johannes would take under this reconstructed devise. The question is, what estate did the testator intend him to take? Whence is derived the authority to make any alteration in the devise as written? Clearly this was a case where the learned court fell into serious error through attempting to construe something which did not call for construction. There was nothing in the devise that called for the application of artificial rules in order to discover the testator’s intention. A testator must be allowed to be his own interpreter when he expresses himself in language free from obscurity, and which, as by him employed, conveys a certain definite meaning, to the exclusion of any other. When he succeeds in doing this he has expressed his own meaning, and that the law accepts as the equivalent of intention. That this testator so succeeded in this particular devise admits of no question. In grammatical construction the devise is *612entirely correct; and it is so definite in expression and terms that but one meaning can be derived from it. It points unmistakably to an alternative gift, and with equal certitude to the intended alternate beneficiary. Why, then, shall there be an exposition of the devise contrary to the words used? That it is modified or changed in any way by subsequent reference to it in the will, cannot be pretended, for it is not once referred to. That it conflicts with any general scheme which can be derived from the will cannot be urged, for there is not a single provision dependent upon it, or which cannot be enforced concurrently with it. Were the devise uncertain because of ambiguity in, some of the words used, it is quite possible that sufficient could be found in other parts of the will to resolve the doubt; but, entirely intelligible and complete in itself, no borrowed light is needed for any purpose in connection with it. Ex vi termini the devise imports a substitutional gift, to provide against a possible failure of Johannes to take. This is peculiarly the case where the word “or” is interposed between the name of the devisee and words of purchase descriptive of the alternative. “The simplest form of a substitutional gift is effected by the use of the word ‘or/ which is usually construed as implying substitution. But in order for a gift to be substitutional, the legatees seeking to put themselves in the place of the deceased legatee, must show that they take by purchase and not by descent:” 30 Am. & Eng. Ency. of Law, 812. Our own case of Gilmor’s Est., 154 Pa. 523, besides being explicit and directly in point, makes such full reference to the authorities that other citation here is unnecessary. Cases where the word “ or,” followed by words of imitation such as heirs or heirs of the body, has been changed into “ and,” are not infrequent-; but our attention has not been called to a single one where the change was made when the words following the disjunctive were words of purchase. We do not say that the change may not be proper in any such case; but the reasons justifying it must not only be found in other parts of the will, but they must be positively compelling to the common understanding. It is argued that it was no part of testator’s purpose to provide against the *613lapse of the devise to Johannes because it was wholly unnecessary to do so, since under the Act of April 8,1833, P. L. 249, there could have been no lapse. Whether a lapse could or could not have occurred, is of no consequence so far as enabling us to understand what was in the testator’s mind when he wrote his will. How are we to know what his understanding of the act of 1833 was? and how would it help us if we did know? The argument assumes that testator had the act in his mind when he wrote the will, and knew that it prevented lapsing. From this assumption the deduction is made that he must have used the word “ or ” inadvertently or by mistake. The non sequitur here is apparent. The argument proceeds, “If the word ‘or’ was not intended to prevent a lapse, it must unquestionably have been intended for some other purpose; and if the testator was not providing against a lapse, what other purpose could have been intended than to vest a fee simple in the children?” Such clearly was his purpose; but only in the event of the parent’s failure to take. The same estate that the parent would -take if he survived, was given his children in the event of failure. We have indicated quite enough to show on what feeble basis the court relied for its justification in changing the language 'of this devise. At best it is pure conjecture. In construing a will the rule requires that it be read in the ordinary and grammatical sense of the words employed,. unless some obvious absurdity or some repugnance or inconsistency with the declared, intention of the testator, as extracted from the whole will, should follow from so reading it. Where this occurs a con.struction may be adopted avoiding these consequences. Words and limitations may be supplied or rejected when warranted by the immediate context or the general scheme of the •will, but not merely on a conjectural hypothesis of the testator’s intention however reasonable, in opposition to the plain and obvious sense of the instrument. It is the expressed intention that governs. Here it is expressed in no uncertain way, and to give it effect this judgment must be reversed. It is so ordered, and judgment is now entered for the defendant.