Loomis, J.
This is a petition by the administrator with the will annexed, for advice as to the construction of the will pf Stephen H. Heath. The testator died in 1810, leaving the will in question, which was executed in 1808. By it he directs that his personal estate, except certain articles given his widow, be put into a form in which it shall produce interest, and that one fourth of this annual interest be paid to his widow for her life, one hundred dollars to each of his four sons and seventy-five dollars to each of his seven, daughters, and that the remainder of the interest be divided between his sons and daughters in proportion to the shares already given them. The will then provides as follows:—“ And in case any one or more of my said children shall die without a lawful heir or heirs, his, *221her or their part to he divided between the surviving children in proportion bequeathed to them as aforesaid; and at the decease of my said children my will is that the principal of said personal estate be divided as follows:— that the first of my said children that shall die, and others until the last of them shall decease, leaving lawful heirs, they severally shall receive the interest of their ancestor until the last of my said children be deceased; and at the '' death of the last of my said children the whole to be equally divided between the lawful heirs of my said children according to the number of their heirs as shall then survive collectively.”
The last of the children of the testator has recently died, and the principal of the fund is now to be distributed. The heirs entitled to it under any construction of the will are very numerous, and are widely scattered, making it a matter of great difficulty, and hardly of possibility, to trace them all.
Two questions are presented by the petition—one as to the rule for the division of the annual income of the property, the other as to that for the final division of the property.
As to the division of the income—we think it clear that upon the death of any one of the children of the testator, leaving children, these children would take the parent’s share of the income, and so of any later generation; in other words, that the share of the income of any child of the testator leaving children, would descend to his or her representatives of whatever generation, until the death of the last of the children left by the testator. The per stirpes rule we think clearly applicable to the income until the time for final distribution of the property. The- administrator has adopted tins rule in paying over the income, and we think correctly.
The other question, with regard to the rule for the final distribution of the property, is one of much more difficulty.
It is clear that by “ the lawful heirs of my said children,” the testator meant the issue of the several children, and of whatever generation.. If this were all it would be an *222entirely clear case that the per stirpes rule would govern the distribution. The remaining words however create a serious doubt—“ according to the number of their heirs as shall then survive collectively.” What do these words mean ? In the confused and inaccurate expressions of which the will is full' it is very doubtful whether they had any meaning to the testator, but so far as we can give them any meaning they seem to indicate an intention to have an enumeration of the heirs then surviving. As this enumeration would be necessary in case of a per capita division and not necessary in case of a per stirpes division, the language would seem to favor a per capita distribution. The word “ collectively ” also somewhat aids this construction, as it indicates that the testator had in mind the heirs of his children as a whole, and an equal division among them by numbers, and not an unequal division by representation. On the other hand however, the ascertainment of all the heirs would be necessary to a complete distribution under any rule; and under any rule the heirs then living would alone take, whether they took per capita or by representation; while the direction of an equal division has been held to be as applicable to a per stirpes as to a per capita division. Raymond v. Hillhouse, 45 Conn., 473. The language is not therefore decisive of an intention to distribute the property per capita, while there are, we think, overbalancing considerations in favor of the per stirpes rule. In the first place the testator had, we think clearly, although in the same awkwardness of language, given the annual income during the long period of the lives of his children or some of them, to his children and their representatives, and it is difficult to see why he should have adopted- a different rule as to the division of the principal. In the next place the per stirpes rule is one of much more easy practical application. The testator must have exj>ected that, leaving so large a number of children, his descendants living at the time, presumably as it has proved very remote, when the last of his children should die, would be very numerous and widely scattered, and consequently very difficult of *223ascertainment, if indeed it would be possible to ascertain them all. Yet this complete ascertainment and enumeration of the heirs, however numerous and however widely scattered, would be an unavoidable preliminary of the distribution of the property. The share of no one heir could be determined until the whole number of the participants in , the distribution was known. The property would in all probability be locked up at the very outset and a distribution of it indefinitely delayed. If the language of the will were clear the difficulty of carrying it out could not affect its construction, but while it is obscure it is a legitimate consideration, inasmuch as it would be presumed that a testator intended a construction that could be carried out, as against one that could not, or could be only at great trouble and expense. Under the per stirpes principle the non-ascertainment of some of the heirs can not affect the shares of any others not of the immediate- division or subdivision to which the undiscovered heirs may belong. A further consideration in favor of the per stirpes rule is, that this rule has for two centuries commended itself to the judgment of the community as one of justice, and has been and is the rule applied by the law in case of intestate estate. In these circumstances this rule will be applied in the construction of a will where the language of the will leaves the intent of the testator in serious doubt. If we should regard the language here used as so utterly obscure that no conclusion could be reached as to its meaning, we should treat the will as so far void for uncertainty, and the property given by it would become intestate estate. It is arriving at the same practical result to give the will a per stirpes construction. This rule has been repeatedly recognized by this court. Thus, in Lyon v. Acker, 33 Conn., 224, the court says:—“We think it a sound rule that when a devise or legacy is given to heirs or their representatives courts will apply the general principles governing the descent of estates, unless a contrary intention appears. The statute of distribution governs in all cases where there is no will; and where there is one, and the testator’s intention is *224in doubt, tbe statute is a safe guide.” See also Raymond v. Hillhouse, 45 Conn., 474. Tbe same rule has been adopted in Pennsylvania. Fissel’s Appeal, 27 Penn. St., 55; Minter's Appeal, 40 id., 111.
We conclude therefore that the final distribution of the property is to be made per stirpes.
In this opinion the other judges concurred; except Pabk, C. J., who inclined to the opinion that the will might be construed as intending a per stirpes distribution among the several branches of the family of which each of the testator’s children was the head, and a per capita division among the heirs of whatever number or degree making up such branch.