In its initial and, we think, fundamental, aspect the case at bar is simple. A transfers property to B in trust to pay its income to C for life. The trust is irrevocable and A retains no control over the income. Who, then, must pay the tax on that income ? The statute, arbitrarily, perhaps, but explicity, taxes it to C, the beneficiary, or B, the trustee, as the case may be, but never to A, the grantor, 26 U.S.C.A. §§ 161, 162, cf. §§ 166, 167. That would be the short and easy answer were it not for certain factual complications in the establishment of the trust- — -the prior marriage of A and C, their contemporaneous separation agreement, and their subsequent divorce. These lead to a consideration of not only the pragmatic command of the statute, but also the contrary implications of a word used in its interpretation by the ultimate authority. For the Supreme Court has declared that where irrevocable trust income is used to discharge an “obligation” of the settlor, operation of the statute is not intended, and the settlor must bear the tax, Douglas v. Willcuts, 296 U.S. 1, 56 S. Ct. 59, 80 L.Ed. 3, 101 A.L.R. 391.
Just where the “obligation” ends and the intendment of the statute begins has been the subject of much learned explanation (and/or conjecture), see, 4 Paul & Mer-tens, Law of Federal Income Taxation § 34.168 (1939 Supp.); Magill, Taxable Income 207, 208; 52 Harvard Law Review 480 (note) ; 87 University of Pennsylvania Law Review 337 (note) ; Bloomenthal, Income Tax Aspects of Alimony Trusts, 17 Taxes 455; Hines, Tax Aspects of Property Settlement Agreements, 12 Southern California Law Review 386; Paul, Five Years With Douglas v. Willcuts, 53 Harvard Law Review 1. The dividing line may soon receive further binding delineation, see Helvering v. Fitch, certiorari granted October 9, 1939, 60 S.Ct. 103, 84 L. Ed. — reported below, 8 Cir., 103 F.2d 702; Helvering v. Leonard, petition for certio-rari filed, 60 S.Ct. 511, 84 L.Ed.-reported below, 2 Cir., 105 F.2d 900. However, we think an accurate demarcation in the instant cause may be accomplished by testing our peculiar circumstance against the two main factors which pervade the controlling decisions.
The first is the nature of the “obligation” ruled upon. In all cases it has been “preexisting”, i e., alimony decreeable by court, Douglas v. Willcuts, above cited, the support of minor children, Helvering v. Schweitzer, 296 U.S. 551, 56 S.Ct. 304, 80 L.Ed. 389, a pre-existing debt, Helvering v. Blumenthal, 296 U.S. 552, 56 S.Ct. 305, 80 L.Ed. 390. The second, cognate to the first, is the element of tax avoidance.
As Professor Magill puts it:
“* * * the debtor has created a trust, the income of which is to be used to pay off the principal and interest of his obligations, remainder to himself or persons of his choice. If he paid off the debts directly out of his own income, he would have no deduction for the sums paid in discharge of principal. By use of the trust device he seeks to reduce his own taxable income by the amount of the income from the corpus of the trust; arid thus in effect to obtain a partial deduction from his income for the amounts used to discharge his debts.
“Had the husband been required to pay the same amount of alimony directly, he would have been taxable upon the total amount of his net. income with no deduction therefor. His income tax liability should not be changed by his creation of a trust to take care of his obligation for him.”
Magill, above cited, pp. 213, 243.
A like possibility of sidestepping the non-deductibility of a capital outlay inheres, of course, in the application of trust income to the support of wives and minor children. What, therefore, is the “obligation”, if any, and the avoidance, if any, implicit in the case at bar?
The income sought to be taxed arises from an irrevocable trust established in 1921 by petitioner-appellant for his wife, both residents of Pennsylvania. The trust is part of a separation agreement, which recites the marital differences of the parties, their living apart, and as consideration for the transfer in trust, the wife’s release of her dower etc. rights. A fixed portion of trust income or if, need be, principal is payable in instalments to her for her support for life, subject to reduction if she remarries, while the balance of income, if any, goes to her husband. On her death the remaining principal reverts to him. As long as she lives, however, he must, if the total income falls below her portion, assign certain securities (comprising his interest in three decedents’ estates) to the trust in order to make up the deficiency. The *986specific payment which the Board has found to be taxable to the husband is one made in 1934, thirteen years from the inception of the trust after the wife had obtained an absolute divorce and remarried (twice).
The divorce occurred in 1921 twenty days after the separation agreement was reached. At that time the statutes of Pennsylvania allowed alimony pendente lite, and on limited divorce (a.m.t.), see 23 P.S. §§ 46, 47. Provision for alimony on absolute divorce was made in but two cases: .first, where the husband divorces the wife for “cruel and barbarous treatment or indignities to his person”, 1895, P. L. 3081 *; second, where the divorced spouse, husband or wife, is insane, 1905, P.L. 211, 212.2 As a consequence the courts were without power to decree alimony in any other circumstance to one freed from the bonds of matrimony. The theoretically indispensable sanction of express statutory authority was lacking, 17 Am.Jur. § 513; 2 Schouler on Marriage, Divorce, Separation, and Domestic Relations § 1757; 2 Vernier, American Family Laws § 104, p. 255. The rationale of that requirement has been tersely stated: “Divorces a vinculo matrimonii were not decreed in ecclesiastical courts. Hence alimony was ’granted by them only as an incident to a divorce a mensa et thoro * *. Except where the common law rule has been changed by statute, alimony cannot be awarded upon a divorce a vinculo matrimonii or sentence of nullity.” Moore v. Moore, 64 Pa.Super. 192, 194. So the lawgivers of Pennsylvania have refused to advance a step beyond the medieval notion of alimony as an incident to limited divorce, see Shoop’s Appeal, 34 Pa. 233. At the same time they have recognized the modern concept of absolute divorce. But no matter how great the wickedness of the husband— be he bigamist, bully, philanderer, or worse, see 23 P.S. § 10 — his innocent wife must risk the poorhouse to be rid of him. The miscreant must go on feeding her only if she is of a mind to quit his bed and board alone, and, correspondingly, if his villainy is less, see 23 P.S. § 11. We may say that this rather unchivalrous anomaly is unique among the forty-eight states, see 2 Vernier, above cited, § 104, p. 263, § 105, Table 53, p. 272. Nevertheless, it is the law, Moore v. Moore, above cited, and see Moore v. Moore, 3 Cir., 255 F. 497; Myers v. Myers, 17 Pa. Dist. & Co. R. 236; Henry Oliver Rea v. Commissioner of Internal Revenue, 35 B.T.A. 1132; Robert Glendin-ning v. Commissioner of Internal Revenue, 36 B.T.A. 486.
By hypothesis,' then the “obligation” here can have nothing to do with alimony, either decreed or decreeable. The conceded fact that the wife obtained the divorce establishes the petitioner-appellant as the offending party and rales out (anomalously again) any application of the divorce statute. It will not do, therefore, to play Hamlet without Hamlet and echo the reasoning of the leading case: “In the present case, the net income of the trust fund, which was paid to the wife under the decree, stands substantially on the same footing as though he had received the income personally and had been required by the decree to make the payment directly.” Douglas v. Willcuts, 296 U.S. 1, 3, 56 S.Ct. 59, 63, 80 L.Ed. 3, 101 A.L.R. 391.
Further, and by the same .token, the common law duty of the husband to support his wife hardly supplies an “obligation”. That duty ended with the divorce decree, 23 P.S. § 55, and it cannot be said with any degree of realism that trust income of $5,000 received in 1934 has any connection with, much less discharges, an obligation to support for three weeks in 1921. See, Blumenthal v. Commissioner, 34 B.T.A. 994,. affirmed 2 Cir., 91 F.2d 1009; Rea v. Commissioner, 35 B.T.A. 1132; Commissioner v. Tuttle, 6 Cir., 89 F.2d 112; Fitch v. Commissioner, 8 Cir., 103 F.2d 702, above cited; Helvering v. Leonard, 2 Cir., 105 F.2d 900, above cited; Fuller v. Helvering, 2 Cir., 105 F.2d 903.
We turn, then, to the more perplexing field of consensual undertakings, exemplified by the pre-existing debt case. The mere fact that the trust was created by an enforceable agreement does not, of course, technically impose any contractual obligation on the settlor with relation to trust income once the trust is set up. And practically speaking, to hold otherwise would be to tax, despite the statute, every trust created by agreement to the settlor, no matter how insignificant the consideration. If the concept of taxable income as a flow of intangible “satisfactions” is to be so drastically applied without statutory *987warrant (compare 26 U.S.C.A. § 167 (a) (3), Burnet v. Wells, 289 U.S. 670, 53 S.Ct. 761, 77 L.Ed. 1439), but rather with the opposite, it should be done by the Congress not the courts.
There remains, however, the contingent obligation to make good income deficiencies, which in 1934 was not discharged but rather prevented from having to be discharged by the income payment sought to be taxed. This circumstance in the shape of a promise or duty to pay was present sub silentio in Douglas v. Willcuts, above cited, and in two recent decisions of this court which also placed the tax on the set-tlor, Alsop v. Commissioner, 3 Cir., 92 F.2d 148, certiorari denied 302 U.S. 767, 58 S.Ct. 480, 82 L.Ed. 595; Glendinning v. Commissioner, 3 Cir., 97 F.2d 51. Its history and importance as a factor in the determination of alternative taxability have been lucidly explained by an outstanding author. See Appendix to this opinion. Nevertheless considerable legal spinning attends either the affirmation or denial of tax incidence upon the settlor-guarantor of trust income. If he is to be taxed on that income it is somewhat difficult to discern what, if anything, he is avoiding. Where the trust is set up to meet the debt, he must be taxed to prevent him, as we have seen, from circumventing the non-deductibility of debt payments. Where, on the other hand, the trust antedates a true guaranty, there is no circumvention of non-deducti-bility, even if the guaranty ripens into a debt. However, if trust and guaranty are roughly contemporaneous in origin, it does not seem unduly cynical to unravel the whole transaction into a continuing obligation to pay a stated amount at future intervals to which payment trust income is applied. So the circumstance is brought within the pre-existing debt criterion, and the settlor is taxed.
Unfortunately for the Commissioner, a similar unravelling of the instant transaction leads to an opposite result. Petitioner-appellant’s credit is not involved, and the ultimate risk of income loss from a diminution of the trust corpus is upon his wife. He did not indirectly promise to pay her an annuity. His direct promise was to create a trust, which he did, and, conditionally, to increase its corpus by a limited amount of securities. The indirect method of achieving the same end would have been simply to set up a larger trust. That of course would have been squarely within the statute and, further, would not have altered his tax position one iota. As ben-ficiary of the surplus income under the hypothetical larger trust, he would be taxed to precisely the same extent as he is now on the surplus income from the actual smaller trust combined with that of the Securities he is under a contingent obligation to transfer to it. Therein, we think, lies the distinction. An ostensible guarantor of trust income must be taxed, or else his tax liability, will not coincide with his true economic status of non-professional annuity writer. Plaintiff-appellant’s tax liability, on the other hand, coincides with his economic position whether or not that position discloses him as the settlor of a large or of a small trust. There is no element of tax avoidance here.
That being so, we find no occasion to extend the precept of Douglas v. Willcuts — -a rule of construction framed to prevent avoidance — beyond the discharge of a statutory common law, or contractual obligation to the non-discharge of the contingent contractual obligation at bar. The benefit to the petitioner-appellant of not having to assign securities to the trust seems to belong among the flow of intangible satisfactions, which we have already mentioned. Its presence, without avoidance, merely shows that it might be desirable to alter the statute. Again, and for more persuasive reasons, it might not. The touchstone of “ability to pay” connotes money rather than fun.
The decision of the Board of Tax Appeals is reversed.
Appendix.
“But although the existence of a legal obligation continuing into the tax year is essential to the husband’s taxability, it seems clear that such an obligation may arise by virtue of the husband’s voluntary assumption of a continuing duty to support just as effectively as by the force of statute or an order of court. If the husband, in contemplation of divorce, sets up a trust by which he personally guarantees a stipulated income to the wife, the trust indenture itself constitutes a continuing obligation which renders the husband taxable upon the income used to discharge that obligation, irrespective of the wife’s right *988to alimony. Relief from an ordinary contractual obligation to pay money constitutes taxable income to the obligor no less than relief from an involuntary obligation imposed by law.
“This argument, although frowned upon in the Board when first advanced by the Government, has since been increasingly favored by both the Board and the courts. Tested against this theory there may be a basis for taxing the settlor even in states which have no provision for alimony. Implicit in the doctrine is the assumption that there is a valid consideration for the contractual guaranty of income assumed by the husband; but the wife, even though not entitled to alimony which might be exchanged for the trust income, may supply such consideration by a release of her dower rights or other rights in jointly owned property.
“This theory, which Unhappily has often been applied without any clear articulation in the decisions themselves, seems a wholly reasonable approach. No logical tax distinction can be drawn between the discharge of a support obligation assumed by a voluntary contract and the discharge of a similar obligation imposed by law; even the latter type of obligation is directly attributable to the voluntary act of the husband in entering into the marriage contract. The contractual theory of the cases under discussion is reminiscent of the early Treasury rulings upon this subject, taxing the husband where the alimony trust was created only as collateral security for the performance of a continuing duty of maintenance. A highly technical argument might be advanced to the effect that so long as the trust income is equal to the annual amount guaranteed by the husband, no actual obligation on the part of the husband has arisen, and therefore no taxable relief from an. obligation has occurred. Still the very absence of such an existing obligation would, even under this view, be due directly to the trust income; and the husband, in being protected from the obligation, has thus received an economic benefit which may fairly be taxed to him.”
Paul, Five Years With Douglas v. Will-cuts, 53 Harvard Law Review 1, 9-12