Atkinson, Justice.

(After stating the foregoing facts.) The dispute in the instant case arises from the fact that the testatrix in item 10(a) of her will uses the language “equally among the grandchildren,” and ends the sentence with the language “per stirpes and not per capita.” The words “per stirpes” and “per capita” both appearing in the same phrase, in the initial approach to the construction of this item, an understanding of the meaning of these terms becomes essential to a proper interpretation of this item. “When issue are said to take per stirpes it is meant that the descendants of a deceased person take the property to which he was entitled or would have been entitled if living. Rotman *522 skey v. Heiss, 86 Med. 633 (39 Atl. 415) ” 32 Words & Phrases (Perm, ed.), p. 459. “By or according to stock or root; by right of representation.” Bouvier’s Law Dictionary; 48 C. J. 808, § 2. Our Code, § 113-902, applies the meaning of this term. “Per stirpes always presupposes an irregularity of relationship.” Payne v. Rosser, 53 Ga. 662, 664. “Persons standing in unequal degrees are allowed to take per stirpes ‘to fulfil,the equity of the statute, which contemplates an equal distribution.’ ” MacLean v. Williams, 116 Ga. 257, 259 (42 S. E. 485, 59 L. R. A. 125). “Per capita” is defined: “By the heads or polls; according to the number of individuals; share and share alike.” 48 C. J. 807, § 2. “‘When descendants take as individuals, and not by right of representation, . . they are said to take per capita.” Bouvier’s Law Dictionary. (Italics ours.) The Code, § 113-902, so applies the meaning of this term. See also Almand v. Whitaker, 113 Ga. 889 (39 S. E. 395).

The plaintiffs in error contend that the word “equally” means an equal division among the sixteen grandchildren, and that the phrase “per stirpes and not per capita” should be disregarded as having no legal meaning. The defendants in error insist that “equally” as here used means an equal distribution among the three different groups of grandchildren of the three named sisters, and that ’“per stirpes and not per capita” has reference to the named sisters. This item of the will concludes with the statement that the fund shall be divided “per stirpes and not per capita.” The plaintiffs in error insist that the previous provision, that the fund “shall be divided equally among the grandchildren” of named sisters, was a provision for a per capita distribution among these grandchildren. To construe the will so as to divide “equally” among the grandchildren, would necessarily require that the last phrase, “per stirpes and not per capita,” be disregarded, as there could be no “equal” division except upon a per capita basis, which basis the will states shall not apply. Where the will uses words which have a well-settled, definite meaning in the law, and there is nothing in the will itself to indicate an intention of the testator that such words should be given any other meaning than that which the law gives them, then it is to be presumed that the intention of the testator was that the words should be construed in the sense that the law would ordinarily construe them. The pri *523 mary consideration in construing wills is the ascertainment of the intention of the testator. Bach will is a law in itself. General rules of construction must necessarily be considered, but previously adjudged cases may be of little authority and even dangerous to apply, and only cases which are in every respect directly in point and agree in every circumstance will afford much aid in determining the testator’s intention. The item now under consideration specifically states that the distribution shall not be on a per capita basis. Considering that the testatrix eliminated any per capita distribution, then it becomes necessary to reconcile the expression that the fund “shall be divided equally among the grandchildren of my . . [named] sisters, . . per stirpes.” Here the term '“equally” can be applied as referring to an equal division among the three different groups of grandchildren of the named sisters. Under such interpretation the words “equally,” “per stirpes,” and “per capita” are each used in the sense that the law would ordinarily construe them. To interpret this item otherwise would require an absolute disregard of the phrase “per stirpes and not per capita.” '“Per stirpes” in its strict sense is used with respect to substituted legatees, that is, legatees who will stand in the place of an original legatee, and does not relate back to ancestors prior to the class that are legatees. It may be that the use of the words “per stirpes” was inaptly applied to the status of the -legatees; but, construing its meaning as being applied to each of the designated three sisters, it takes on a meaning of legal significance. Inasmuch as this item states that the distribution is not to be made per capita (an expression that is clear and unequivocal and has a precise and definite legal meaning), an inapt or inexact use of the words '“per stirpes” would not negative the express intent that the distribution was not to be per capita, especially where the testatrix is devising to three different sets of grandchildren of three named sisters. The clear statement that the distribution is not to be made per capita must, if possible, be given effect. The only other mode of distribution would be per stirpes, and even though the application of the term “per stirpes” was not, in its strict sense, correctly used; yet it denotes a manner of distribution different from a per capita distribution, and is indicative of an intention on the part of the testatrix, to which the courts should give effect, even though the term is not used in its strict legal sense.

*524 We find no decision of the courts of this State where identical language in a will has been construed. In the excellent brief for the plaintiffs in error, we are cited to decisions by the courts of other States where language very similar, though not identical, has been construed to require a per capita distribution. We have examined those decisions, and though the language used is very similar, we do not think that' any is direct authority for a ruling contrary to the interpretation already stated in this opinion. In Ives’ Estate, 161 Misc. 60 (291 N. Y. Supp. 981), it was held that a per capita distribution was required under the following clause: “Upon the decease of my said wife and the termination of her said life use of such property, I then give, devise, and bequeath the corpus or principal thereof to my nephews and nieces, share and share alike per stirpes not per capita.” The decision states, “the words per stirpes not per capita being treated as meaningless and nugatory, and as being inserted by inadvertence.” There, it will be noted that the testator’s brothers and sisters, the parents of the nephews and nieces, were not named or referred to. Cognizance of this fact was noted by the following statement in the decision: “Testator does not refer to the legatees as children or the next of kin of his brothers and sisters. He makes no reference to those who are to receive the fund at the termination of the life-estate in any manner that conveys the thought that they are to take according to stock or root.”

In Title Guaranty & Trust Co., 159 App. Div. 803 (144 N. Y. Supp. 889), the following language was construed to mean a per capita distribution: “I give and bequeath the said sum . . unto the children then living of my sons Charles . . and William, . . and the issue of such as may have died leaving issue then surviving, per stirpes and not per capita.” In the opinion it was stated that, as punctuated, the bequest would go to the “children of the testator’s two sons . . per stirpes and not per capita;” but also that it would be an unusual use of the words “per stirpes.” In ruling that the distribution should be made per capita, the court assumed that a comma was intended between the words “died” and “leaving,” the way in which identical language in another part'of the will was punctuated; and the will as a whole showed an intent to distribute the property-to the grandchildren per capita.

*525 In Foster’s Estate, 144 Misc. 622 (259 N. Y. Supp. 450), the will provided for a '“distribution amongst the heirs of my sister, Mary Rhymus . . and the heirs of William Foster (an uncle) their heirs and assigns forever, per stirpes.” At the death of the life-tenant (Mary Rhymus) all her descendants were dead. The uncle (William Foster) and his two children predeceased the life-tenant, but William Foster’s two children had left children, one having three surviving, and the other one. The court held that these four grandchildren of William Foster, being related in the same degree, should share equally, and that a per stirpes distribution would have applied only to a deceased grandchild leaving children. But here it must be noted that this provision of the will did not provide that the distribution should not be made per capita, as does the will now under consideration. Neither did the will in Duval’s Estate, 145 Misc. 792 (261 N. Y. Supp. 884), state that the distribution was not to be per capita.

In Perlmutter’s Will, 156 Misc. 571 (282 N. Y. Supp. 282), at the termination of a testamentary trust the corpus was to be divided “equally among my surviving children, per stirpes and not per capita.” From an examination of this case, it does not appear that there was any issue as to whether the distribution was to be made per capita or per stirpes. The court held that it should be divided among the children, and, if any should be dead leaving descendants, they would stand in the place of their deceased parent. There the will dealt with only one set of children, and per stirpes could only apply to their children. It is unlike the instant ease, where the distribution was to be made to the grandchildren of three named sisters.

In Camden Safe Deposit & Trust Co. v. MacMullan, 112 N. J. Eq. 574, 577 (165 Atl. 105), a testator, in the event of certain contingencies, gave his estate to eight named nieces, five daughters of one sister, and three of another, share and share alike, per stirpes and not per capita, the lawful issue of any deceased niece taking the share which such deceased niece would have taken and received if she had lived. The court said: “The testator apparently intended to direct an equal division . . among his eight nieces because he named them without reference to the parent. It would seem that if he had intended to divide them into classes, and give one-half of his estate to the three daughters of *526 his sister, Mrs. MacMullan, and the other one-half . . to the five daughters of his sister, Mrs. Simpkins, he would have directed a division among the daughters of these two sisters ‘per stirpes and not per capita/ omitting the phrase ‘share and share alike.’ ”

In another case cited by the plaintiff in error, Petition of Gee, 44 R. I. 132 (115 Atl. 716), the residuary clause provided: “All the rest and residue of my estate I give . . to my said wife and to my sons and daughters . . share and share alike, per stirpes and not per capita.” The wife was a second wife, and not the mother of the five children. It was held that the wife did not take one-half of the residue, but that it should be divided into six parts, the wife taking a one-sixth part. This conclusion was reached primarily by gathering the intention of the testator from the general scheme of the will, which denoted an intention to make an equal distribution among all six legatees.

None of the foregoing decisions from other States dealt with the identical question here presented, as all may be distinguished from the instant case either by the phraseology or by other provisions contained in the various wills. While the tenor of some of those decisions seem, as insisted by the plaintiff in error, to disregard the phrase “per stirpes and not per capita,” yet in the instant case we feel that the better approach, to ascertain the intent of the testatrix, is to give effect to these words by taking cognizance of the phrase “not per capita,” and by giving some effect to the inexact use of the words “per stirpes.” From a perusal of the will as a whole, we are unable to find evidence of an intention on the part of the testatrix contrary to what is above indicated. Aside from the two-fifths interest in the residue, which is here in question, under items 10(b), 10(c), and 10(d), each of the sisters, who are the grandmothers of the parties to this proceeding, was given a life interest in one-fifth of the residue. At the death of each of these sisters, under the provisions of each of the above items, their grandchildren, the grandnieces and grandnephews of the testatrix, took the property in fee. Each group of grandchildren took a separate and distinct one-fifth portion in which their grandmothers had a life interest. So, it is clear that it was the intention of the testatrix that these three-fifths of the corpus were not to go to her grandnephews and grandnieces on a per capita basis, but according to stock. Accordingly, the ruling *527 of the trial judge that the property be divided into three parts, rather than sixteen, and that each set of grandchildren should participate in their respective portions, was a correct interpretation of the will.

At the hearing before the trial judge, the plaintiff in error offered the testimony of Mrs. Jessie D. Dixon for the purpose of showing the circumstances surrounding the testatrix at the time of the execution of the will, which evidence was excluded. The evidence would have shown that the nine grandchildren of Mrs. Dixon were doubly related to the testatrix because the testatrix and Mrs. Dixon, who were sisters, had married brothers; and would also have disclosed that the testatrix had inherited approximately one-third of her estate from her husband, and that one of her grandchildren was named for the husband of the testatrix. The Code, § 113-807, provides: '“When called upon to construe a will, the court may hear parol evidence of the circumstances surrounding the testator at the time of its execution.” The court should have heard this testimony, but we do not think that its exclusion was such error as would require a reversal of the decision construing the will. This evidence would not have authorized the trial judge to have construed the will differently, as the intent of the testatrix was covered by the language in the will.

It was not error to exclude the copy of the will of Mrs. Nellie Dale, and the testimony of Mrs. Merritt Dixon Jr., in explanation thereof, for the reason that knowledge of the contents of this will on the part of Mrs. Lillie D. Dixon, the testatrix in the instant case, was not properly established, being based only upon opinion testimony. Nor did the trial judge err in admitting the testimony of James M. Rogers; the objection interposed being to the whole of the testimony, and some of it being germane and proper.

Judgment affirmed.

All the Justices concur.