— The troublesome portion of this will is the residuary clause which provides payment of income “to my sons in equal shares the children of any son of mine who may be deceased at the time of my death to receive the income which the parent would have received if living. But upon the death of any of my sons who may he living at the time of my death, then I direct the proportion of the income which such son would have received if living to be paid to the children in equal shares of said deceased son, and if said son shall die leaving no children living said proportion of the income which such son of mine would have received if living shall be paid to the widow of said son till her death or re-marriage. Upon *498the death of the Survivor of my sons and their respective wives, I direct that my estate shall be divided among my grandchilren then living per stirpes and not per capita and this Trust shall thereupon cease and determine.”
The auditing judge has ruled correctly that the residuary clause offends- the rule against perpetuities in that the remainders to the grandchildren are not to vest until . . the death of the survivor of my sons and their respective wives”, emphasized in the phrase “grandchildren then living”. While possibly the testatrix intended the substitutionary gifts to the sons’ wives to extend only to such wives as were living and married at testatrix’s death (Solms’ Estate, 253 Pa. 293; James’ Estate, 245 Pa. 118; Wright’s Estate, 284 Pa. 334; Anshutz et al. v. Miller, 81 Pa. 212; Packer’s Estate (No. 2), 246 Pa. 116, and other similar cases), yet there is no authority under which a woman born after testatrix’s death and married to her son Horace could have been eliminated, nor can we find any provision that permits of severability of shares, of events, of parties, or of subject matter under which other interests might be salvaged, as in Whitman’s Estate, 248 Pa. 285, and Kern’s Estate, 296 Pa. 348. See also A. L. I. Restatement of Property, Future Interests, §243. Mr. Gray in “The Rule Against Perpetuities” (4th ed.), p. 369, ch. IX “Separable Limitations”, says:
“Very often, indeed generally, a future contingency which is too remote may in fact happen within the limits prescribed by the Rule Against Perpetuities, and a gift conditioned on such contingency may be put into one of two classes according as the contingency happens or does not happen within those limits; but unless this division into classes is made by the donor, the law will not make it for him, and the gift will be bad altogether.” (Italics supplied.)
*499This case is one of the illustrations of the soundness of this principle, because of the difficulty we would encounter in attempting to separate the interests of the wives of sons married at the date of testatrix’s death from Horace’s interest. Here, the gift over to grandchildren “then living” occurs “upon the death of the survivor of my sons and their respective wives” and is not separated from other possible events, i. e., when the sons die with or without children and whether or not the share of any wife of a son ever actually vests. In addition to the well-selected authorities relied upon by the auditing judge, Lockhart’s Estate, 306 Pa. 394, following Coggins’ Appeal, 124 Pa. 10, shuts off all argument on this point when it holds (p. 401) that “If the gift is to a class and it is void as to any one of the class, it is void as to all ...” Likewise, in Wanamaker’s Estate, 335 Pa. 241, 246, Justice Schaffer, in reversing the action of the lower court which held that remainders were not too remote because they were to vest only at the death of the survivor of grandchildren living at the date of the death of the testator, said:
“We conclude that the confused language used by the testator defers the vesting of the remainder interests until twenty-one years after the date of the death of the last survivor of his grandchildren as a class, irrespective of whether or not the grandchildren were living at his death. The language used does not limit the life estates to grandchildren living at the death of the testator, but to grandchildren generally.”
The reference therein to Hogg’s Estate, 329 Pa. 163, and other cited cases is very pertinent. See also 3 Page on Wills (Liff time ed.), ch. 31, sec. 1252, p. 656, and other authorities cited in adjudication.
In Dean v. Mumford et al., 102 Mich. 510, cited by ex-ceptants, the interests of the three sons were separated, the vesting of the remainders was not dependent upon *500the death of any or all surviving widow or widows, but each share vested in turn upon the death of the particular son and of his wife: “the portion so held in trust . . . shall become the property of and go to the child or children of said sons, severally and their heirs and assigns, forever.” Griscom’s Estate (No. 2), 336 Pa. 422, is distinguished in that the Supreme Court there held that the will expressly provided the valid contingency.
The explanation given by Jessell, M. R. in Miles v. Harford, 12 Ch. D. 691, cited in the footnote on page 379 of the fourth edition of Gray’s Rule Against Perpetuities, is peculiarly appropriate:
“ What I have said is hardly intelligible without an illustration: On a gift to A. for life with a gift over in case he shall have no son who shall attain the age of twenty-five years, the gift over is void for remoteness. On a gift to A. for life, with a gift over if he shall have no son who shall take priest’s orders in the Church of England, the gift over is 'void for remoteness; but a gift superadded, ‘or if he shall have no son’, is valid, and takes effect if he has no son; yet both these events are included in the other event, because a man who has no son certainly never has a son who attains twenty-five or takes priest’s orders in the Church of England, still the alternative event will take effect, because that is the expression.
“The testator, in addition to his expression of a gift over, has also expressed another gift over on another event although included in the first event, but the same Judges who have held that the second gift over will take effect where it is expressed have held that it will not take effect if it is not expressed, that is, if it is really a gift over on the death before attaining twenty-five years or taking priest’s order, although, of course, it must include the case of there being no son. That is what they mean by splitting, they will not split the *501expression by dividing the two events, but when they find two expressions, they give effect to both of them, as if you had struck the other out of the will. That shews it is really a question of words and not an ascertainment of a general intent, because there is no doubt that the man who says that the estate is to go over if A. has no son who attains twenty-five, means it to go over if he has no son at all, it is, as I said before, because he has not expressed the events separately, and for no other reason. ...”
The conclusion that the trust terminated on February 28, 1938, upon the death of the last surviving son is correct, even though there were then living the widows of two sons, which sons were survived by children. The event fixing the termination of the trust is not the date of the death of the last surviving wife of a son, because such event under the will determines the trust only if the son dies without children and such wife takes a life estate by substitution — an event which never happened because all four sons were survived by children.
This conclusion is strengthened by the failure of testatrix to limit the substitutionary gift of income to the children of any son dying after her death, to any particular period, much less for the life or lives of any surviving wife of any son of testatrix. The language she employed clearly makes such gifts over of income distributable per stirpes- and not pur autre vie: Nixon’s Estate, 306 Pa. 261. When then do such shares of income cease and merge into principal shares? Certainly they should not be required to await an event so unrelated to its associated provisions as the death of named contingent beneficiaries whose gifts never vested.
The confusion involved in this question is no doubt responsible for the trust continuing beyond the audit of the 1938 account before the writer of this opinion, at which time the trust was not declared terminated *502because it was represented to the court that the trust “continued under an agreement of all parties in interest”. ■
We likewise agree with the adjudication in its conclusion that the principal of the estate comprising assets other than unconverted real estate is distributable under the authority of agreement of the parties entered into in 1932 upon the pattern laid down in the will and not under the intestate laws. It is evident that the agreement of 1932 was entered into for the purpose of having a unified control of the sale and conveyance of unconverted real estate, of providing a system of substitutionary limitations for the distribution of income in which particular the will was lacking, principal of unconverted real estate to follow the distribution of income thus outlined, and finally a distribution of personalty principal, which is the subject matter of the present accounting. All these purposes were to be subserved for the benefit of the signators from “the entire right, title and interest of testatrix’s sons, their widows and her grandchildren to the said residuary estate of the said testatrix whether under the will above mentioned or by intestacy or otherwise”, which interest “has vested in the parties hereto; all of whom are sui juris. . . .” We regard as groundless the exception that this is not a binding agreement under the authority of Wanamaker’s Estate, supra, and of Straw-bridge’s Estate, 322 Pa. 406. It is true that the parties in Wanamaker’s Estate specifically stated in their agreement that it was to prevail in the event that the provisions of the will to which it referred violated the rule against perpetuities. The absence of such specific language in the instant instrument is readily supplied by the specified eventuality of the parties deriving their interests “under the will ... by intestacy or otherwise”. The last two contingencies “by intestacy or otherwise” were clearly meant to constitute the only two alternative manners of their inheriting in the *503event the will violated the rule. Conner’s Estate (No. 2), 318 Pa. 150, is authority for the sufficiency of the word “same” in the granting clause of the agreement to constitute an incorporation by reference of the entire trust res. In passing, it is not clear to us, however, for what reason, other than the one mentioned later herein, all of testatrix’s grandchildren were asked to and did enter into this agreement by which, under the construction given it by the auditing judge, they contracted that if they died prior to Jacob’s death their estates and their children would be disinherited.
Exceptions were filed to the ruling of the auditing judge that the interest of Rachel McVicker, who did not execute the agreement, now exists in her representatives. Following the argument, all interested parties executed releases which have been filed of record, to the end that Mrs. McVicker’s interest in the estate is effectually waived. Therefore, this question disappears from the case, and the exceptions covering it are with the approval of the auditing judge sustained pro forma.
The adjudication rejects the suggestion that the gift in remainder to “my grandchildren then living per stirpes and not per capita” includes testatrix’s great-grandchildren, children of deceased grandchildren, relying upon the authority of Williamson’s Estate, 82 Pa. Superior Ct. 444, where the clause in remainder read “among the grandchildren of his other brothers and sisters then living per capita and not per stirpes”. There it was held that the phrase “per capita and not per stirpes” referred only to the quantum of the share of every grandchild and that “grandchildren” did not include great-grandchildren. It is important to observe that the language in the cited case is the reverse of that used here — “per capita and not per stirpes” as opposed to “per stirpes and not per capita”.
The general principle of law involved is that the word “children” has a sense and meaning which will *504not be extended to include great-grandchildren: Dickinson v. Lee, 4 Watts 82; Hallowed et al. v. Phipps et al., 2 Wharton 376; Castner’s Appeal, 88 Pa. 478; Page’s Estate, 227 Pa. 288. In the case of “grandchildren” being extended to include “great-grandchildren”, the auditing judge apparently was of opinion that Williamson’s Estate, supra, applies the same strict rule in the case of “grandchildren”. However, a different approach is made to this terip in Pemberton v. Parke, et al., 5 Binney 601, 610, where it is said that “Grandchildren are words of equivocal import, and may or may not include great-grandchildren, according to the sense in which they may have been used ...” In Hallowell et al. v. Phipps et al., supra, the reason for the limited definition of “children” is stated to be that testator’s children are always known to him and they enjoy his care and guidance. By the same token, this testatrix in making her will in 1894 had as little opportunity of knowing who her grandchildren would be as she had of knowing her great-grandchildren.
The exceptions to the foregoing general principle of law are as Gibson, C. J., stated in Dickinson v. Lee, supra: “. . . where the word is used evidently as coextensive with issue, which is a word of very general import; and where there are no children literally to answer the description, and then grandchildren are let in ut res magis valeat qumn pereat”; and in Mc-Glensey’s Estate, 37 Pa. Superior Ct. 514, where it is the clear intention of testator as shown by the whole will that the class be extended.
An examination of the will satisfies us that its language comes within two of these three exceptions.
In attempting to analyze testatrix’s intent, certain things are clear. The phraseology she employed in this clause, “grandchildren then living”, indicates intent to exclude the estates of deceased grandchildren, and in adding the clause “per stirpes and not per capita” she *505indicates in a primary sense that the beneficiaries are to take by representation and not equally. The addition of the latter phrase was essential, since without it the grandchildren would have taken per capita: Brundage’s Estate, 36 Pa. Superior Ct. 211. In passing it is to be observed, therefore, that in Williamson’s Estate “per capita and not per stirpes” must be regarded as surplusage.
While the phrase “per stirpes, and not per capita” conveys the sense above pointed out, nevertheless, it has another sense in that it includes words which have a distinctly technical meaning and which, therefore, brand the clause as equivocal.
There is an imponderable principle deeply engraved in our law of inheritance, judicially as well as legislatively expressed, that lineal descent shall be presumed, excluding none in the line, and in order that none be excluded childrén take by representation the share of their deceased parent. This presumption in favor of the general rules of inheritance must prevail where-ever possible: Mayhew’s Estate, 307 Pa. 84; Windt’s Estate, 110 Pa. Superior Ct. 124; Hoch’s Estate, 154 Pa. 417; Conner’s Estate (No. 2), supra.
The same principle cannot be said to apply in favor of collateral heirs, which constituted the class of beneficiaries in Williamson’s Estate, supra. The limitations upon representation by collaterals are expressly designated in the Intestate Act, under which the extent of representation has been construed to be the grand.children of brothers and sisters and the children of uncles and aunts: Miles’ Estate, 272 Pa. 329. On the other hand, there is no limitation upon the class or classes of lineals who may inherit under the same Iaws.-
Whenever it appears that the intention of a testator is to send shares down in the line of descent, then grandchildren and remote descendants take as representatives of their deceased ancestor. This principle is *506well illustrated in Campbell’s Estate, 202 Pa. 459, which most nearly approaches the facts of our case. There a bequest was given to three daughters for life, remainder to their children, but if one died without children then to surviving brothers and sisters of such of them as might then be dead. When a daughter died leaving grandchildren, but no children, it was held that such grandchildren, great-grandchildren of the testator, were entitled. The court said (p. 462) what is peculiarly applicable here:
“The manifest general intent of his will was to send the shares of each of his daughters respectively down in the line of her and his blood per stirpes. If he had said ‘decease without leaving any issue surviving/ it would have been beyond question and the plain intent of this whole provision in his will shows that he used the word ‘children’ in the sense of issue, a sense which as remarked by the learned auditing judge, is sanctioned by the lexicographers and by the decisions in Haldeman v. Haldeman, 40 Pa. 29, and Potts v. Cline, 174 Pa. 513.” So it is clear to us that, when this testatrix said “grandchildren then living per stirpes”, she intended a like devolution.
Further authority is contained in the A. L. I. Restatement of Property which states the rule that “children” excludes more remote descendants, and the exceptions thereto.
Section 285 . . . “(2) The following are the more frequently encountered factors tending to establish the existence of the‘contrary intent of the conveyor/ . . . :
“(d) the conveyor effectively manifests an intent for stirpital distribution and such intent is more completely accomplished by a broader inclusiveness of the term ‘children’.”
In the comment on this subsection (2), clause (d), we find the following:
“k. Forms of limitation commonly within subsection (2), Clause (d). When the language is unmis*507takably explicit in directing a distribution in which the offspring of a deceased person represents such deceased person, no one can successfully urge the exclusion of grandchildren or more remote descendants of the designated parent. Such explicitness usually takes the form of either the addition of the phrase ‘per stirpes’ to a limitation in favor of the ‘children of B,’ or the inclusion of detailed substitutional dispositions, in terms made applicable to the share of each child. . . . Such cases are seldom litigated as to the inclusiveness of the term ‘children’.”
Another factor in the instant will which prompts this construction is the position in which the phrase ‘per stirpes and not per capita” appears, i. e., immediately following the designation of the class that is to take. There usually appear in this position words of limitation expanding or restricting the estate first given or providing for substitution. That is the sequence of ideas in the remainder clause of this will and, therefore, the phrase “per stirpes and not per capita” appearing in this position is clearly open to the connotation of “issue”.
The construction given in the adjudication excluding great-grandchildren contradicts the idea of stirpital distribution contained in the word ‘per stirpes”, and is inharmonious with testatrix’s scheme of distribution.
“Where ambiguous or contradictory expressions appear in a will, the law adheres as closely as possible to the general rules of inheritance”: Windt’s Estate, 110 Pa. Superior Ct. 124, syllabus.
In Conner’s Estate (No. 2), supra, it was held that a scheme of distribution unusual in the law requires plain and unequivocal language to establish it. We feel that the scheme of distribution which the adjudication attributes to this testatrix is unusual — it is not published in plain and unequivocal terms such *508as “grandchildren being children of my children” or “grandchildren then living excluding great-grandchildren”.
Hogg’s Estate, supra, is to be distinguished. There the question was, not the exclusion of great-grandchildren, but whether in a single class of grandchildren some should be excluded because their deceased parents had already received an intestate share. “Per stirpes” as used by this testatrix is to be given its ordinary and correct meaning, as defined in Shoch’s Estate (No. 2), 271 Pa. 165, 167:
“. . . ‘taking by representation’ . . . ‘the descendants of a deceased person take the property to which he was entitled or would have been entitled if living . . . The expression “per stirpes” means to take by stock or through a common source, and, when applied to estates, signifies that the particular descendants shall take among themselves the share of their deceased parent,’ . . .”
“Per stirpes” means not only a taking by families as in Hogg’s Estate, but as well a taking by representation and the inclusion of the whole line of descent.
In conclusion, we find that the application of the authority of Williamson’s Estate, supra, relied upon by the auditing judge, is completely nullified, because first the phrase “per capita and not per stirpes” excluded all consideration of the principles herein employed, since the words “not per stirpes” expressly .negatived all possibility of stirpital distribution; second, the phrase “per capita and not per stirpes” in Williamson’s Estate was as heretofore observed surplusage; third, the word “grandchildren” therein employed applied to collaterals of testatrix, whereas “grandchildren” in the present situation involves lineals of testatrix.
Therefore, we conclude that the testatrix expressly and impliedly intended that the remainder clause of *509her will express the meaning that the words “grandchildren” include “great-grandchildren”, and that the grandchildren and the great-grandchildren all share per stirpes, that is through the fathers of the grandchildren, the testatrix’s sons. Incidentally, we believe that this conclusion was shared by the parties when they signed the 1932 agreement; it is the only reason we conceive why they would have risked disinheriting their children by signing the agreement which eliminated all signators who would predecease Jacob Disston.
The auditing judge struck from principal account certain items reflecting the balance of advancements from personalty for real estate improvements made prior to 1932, which items the 1932 agreement did not specifically allocate to real estate or to personalty, and which may be distributable in different channels. These items are yet to be liquidated and hence are not before the court. Further, their distribution may possibly take a different course than that taken by the present fund, and finally the auditing judge’s action was directed without prejudice to the rights of the parties. Under one theory the loans thus made should be the subject of reimbursement and, therefore, constitute a charge against the realty, while under the contrary one they became real estate when advanced and must so remain. The auditing judge had no power under the circumstances to declare that the stricken items constitute charges against the realty: Grothe’s Estate, 237 Pa. 262; Clark’s Estate, 26 Dist. R. 1046.
Such exceptions as are supported by this opinion are sustained, those not so supported are dismissed, and the adjudication thus modified is confirmed.