— We find little to add to the auditing judge’s succinct and correct conclusion.
The exceptants’ argument is that only those grandchildren whose parents were named in the will should inherit, and that the intervening intestacy was not anticipated by testator.
Both rules of construction and arguments on probabilities must give way before positive language, and we are in accord with the auditing judge when he holds that the word “grandchildren” obviously means all grandchildren. There is no ambiguity in it, and it is quite clear that if there were no living children of testator’s four children named in his will the two children of testator’s afterborn children would fit the description.
An heir cannot be disinherited except by express words or by necessary implication. In Robison’s Estate, 266 Pa. 251, 254, there was a gift to “children”, and it was attempted to exclude testator’s son by a former wife from whom he had been divorced. Testator and this son had been estranged for 22 years. The court said:
“We find no express words of disinheritance in the will, but, on the contrary, it appears petitioner is clearly within the class created by the residuary gifts to testa*221tor’s children. Nor do we find in the will, or circumstances under which it was made, anything indicating, or that can be construed as, an exclusion of petitioner by necessary implication. The use of the word ‘children’ necessarily implies all children and an intention to treat all alike: Hoch’s Est., 154 Pa. 417.”
In Luce et al. v. Harris, 79 Pa. 432, testator devised land to his son and daughter-in-law, Jacob and Eliza, for life, and to “their children at their death.” Eliza was Jacob’s second wife, and he had two daughters by his first wife. The court said at page 435:
“In order to confine the generality of the words ‘their children’, to those born of the second marriage and thus disinherit those born of the first marriage, such intention must clearly appear from the face of the will. It must be manifest by its express terms, or by necessary implication. Neither exists in this case.”
This testator saw fit to make his will in 1855, “being about to travel abroad”. Having survived the trip and added three more children to his family, he failed to change his will, as he could easily have done. We know of no case which holds that the law which lets in after-born children exhausts the share which their line of descent may take under a provision in the will whose description fits their issue. We are primarily concerned with what testator said, since it is in clear language.
In Ludwick’s Estate, 269 Pa. 365, 371, the court said:
“We might guess as to his intent, but, no matter which way we conclude, we would do violence to the language actually used, and this we are not permitted to do, since the ‘question in expounding a will is not what the testator meant but what is the meaning of his words’ ”.
The words “per stirpes” in the gift to the grandchildren, which is a gift to a class, determine the shares the grandchildren were to inherit and do not indicate a discrimination between the ancestors from whom they were to inherit.
*222We do not feel that Miller’s Estate, 323 Pa. 9, alters the situation. It is quite as consistent for us to believe that when he wrote his will and used the word “grandchildren” testator contemplated the possibility of having more children as it is for us to believe that when he wrote his will and used that word he contemplated having none, or that he meant to discriminate against unborn children merely by naming living ones.
If it were not for the statute which let in afterborn children and so operated against the will, no one would doubt that the natural meaning of the disputed word when he used it was intended to include all grandchildren. In that event, the hard case which appeals to our sympathies would influence us to read the word in its natural sense. It cannot change that natural meaning merely because the hard case now works the other way.
The exceptions are dismissed and the account is confirmed absolutely.