COFFIN, Circuit Judge.

This appeal presents the question whether the proceeds of a life insurance policy on decedent’s life are properly includable in the gross estate of the decedent by reason of the alleged possession at his death of “any of the incidents of ownership, exercisable either alone or in conjunction with any other person”, under Section 2042 of the Internal Revenue Code of 1954, 26 U.S.C. § 2042. 1

The Commissioner of Internal Revenue having included the proceeds of an insurance policy on the life of Holton W. Horton (decedent) in his gross estate and the sum of $14,185.85 in federal estate taxes and $1,004.67 in interest having been paid, the plaintiffs, co-executors under his will, made timely claim for refund and brought this action for recovery under 28 U.S.C. § 1346(a), alleging that such sums were erroneously assessed. The matter was submitted to the district court upon an agreed statement of facts and depositions. The district court found for the plaintiffs, D.R.I., 1965, 241 F.Supp. 586, and the government appeals.

The facts, undisputed, are of two kinds: “intent facts” — those relating to the conduct and understanding of the insured and his father, who was the instigator, premium payer, and primary beneficiary of the policy; and the “policy facts” — those revealed by the insurance contract itself.

Decedent’s father, Charles A. Horton, was a textile executive, a prominent businessman in his community, and, according to the testimony, “a man with strong convictions and vigorous action”. Charles and his wife, Louise, had two sons, decedent and A. Trowbridge Horton. In 1924, when decedent was 18 and Trowbridge 19, their father purchased an insurance policy on the life of each boy from Massachusetts Mutual Life Insurance Company. The policies were identical, each having the face amount of $50,000, the proceeds being payable to Charles and Louise, equally, or to the survivor.

Charles Horton’s purpose was to assure that funds would be available for his wife, should he and either son die. Charles kept the policies in his safe deposit box and paid all premiums throughout his life. Under the policies, however, the right to change beneficiaries had been reserved to the sons. In January, 1952, the boys’ mother, Louise, died. In March, 1952, Charles told each of his sons to go to the insurance company’s office and sign a change of beneficiary form. The amendment executed by decedent named his father as primary beneficiary, with decedent’s wife, brother, and the executors or administrators of the last survivor being the successive beneficiaries. After this amendment, decedent continued to retain the right to make further changes, but none was made. Decedent died on April 1, 1958, survived by his wife and father. His father died on October 2, 1961.

The father, Charles, regarded the policies as belonging to him, saying at one point that it would be “out of the question” for the sons to claim them. Decedent’s brother never discussed the policies with his father, never asked for a loan based on the policies, obediently *9 signed the change of beneficiary form at his father’s request, and considered the policy on his life as the property of his father. Decedent’s widow recalled only that decedent had once told her that his father had a policy on himself and his brother but that “in no way did it mean anything to us or would it ever. It was completely his.” She added that her husband, the decedent, had wanted more insurance of his own, but was not able to obtain it.

Coming to what we call “policy facts”, a careful reading of the policy, captioned “Ordinary Life Policy — Convertible”, reveals the following rights, privileges, or powers accorded to the decedent.

—Right to change beneficiary. In the application, an unrestricted change of beneficiary provision was elected by striking out two alternative and more limited provisions. 2 The policy itself indicated reservation of “the right successively to change the beneficiary” by the insertion of typewritten dashes where, otherwise, the word “not” would have been inserted.

—Assignment. No assignment would be recognized until the original assignment, a duplicate, or a certified copy was filed with the company. The company did not assume responsibility for the validity of an assignment.

—Dividends. The insured had the option to have dividends paid in cash, used to reduce premiums, used to purchase paid-up additions, or accumulate subject to withdrawal on demand.

—Loans. On condition that the unlimited right to change the beneficiary was reserved, as in this case, the company would “loan on the signature of the insured alone”,

—Survival. Should no beneficiary survive the insured, the proceeds were payable to his executors and administrators.

—Alteration. The policy could be altered only on the written request of the insured and of “other parties in interest”.

—Discharge of company’s obligations. The company would not be responsible for the conduct of any trustee or for the determination of the identity or rights of beneficiaries. Payment at the direction of a trustee or in good faith to a benficiary would discharge the company of its contractual obligations. Beneficiaries were advised by the policy that they need hire no firm or person to collect the amount payable under the policy, but that they would save time and expense by writing to the company directly.

The plaintiffs contend that the district court properly held that decedent possessed no incidents of ownership in the policy; that the term “incidents of ownership” refers to the rights of insured or his estate to the economic benefits of the policy; that the question of possession of such incidents is one of fact; that such possession depends upon all relevant facts and circumstances, including the intention of the parties; and that these facts and circumstances clearly establish that decedent’s father was the real owner of the policy, while decedent was merely the nominal owner, having no real economic interest in it.

The government asserts that, as a matter of law, the facts bring this case squarely within the reach of Section 2042, as applied by the cases, notwithstanding the evidence as to the intentions and extra-policy circumstances of the parties, and the lack of economic benefit to decedent.

At the outset we are confronted with the issue of the nature of this review. It is undoubtedly true that the question of possession of incidents of ownership of a life insurance policy is *10 one of fact, the plaintiff having the burden of proving non-possession of all. Estate of Piggott v. Commissioner of Internal Revenue, 6 Cir., 1965, 340 F.2d 829; Fried v. Granger, W.D.Pa., 1952, 105 F.Supp. 564, affirmed, 3 Cir., 1953, 202 F.2d 150; Hall v. Wheeler, D.Me., 1959, 174 F.Supp. 418; Estate of Collino v. Commissioner, 1956, 25 T.C. 1026. But where all of the evidentiary facts appear, we are faced with a question of law not of fact. Were we to proceed otherwise, cases presenting identical or closely similar facts in this technical and complex field could be decided oppositely, to the disadvantage of equitable tax administration.

Taking the subsidiary facts as presented to the district court, we differ with its conclusion that “the decedent’s father was actually the real owner of the various incidents of ownership in said policy”. But in differing we recognize that early holdings and occasional dicta, early and late, have invited litigation. This is the kind of case where the government enters, appearing to seek its pound of flesh on the basis of petty technicality, while the taxpayer’s decedent generally appears as a person who had very little to do with the insurance policy which is causing so much trouble to his estate. If such hard cases have not made bad law, they have at least made bad dicta. Although the Supreme Court has recently (indeed, since the principal hearing of this case before the district court) spoken strongly and succinctly on this issue in Commissioner of Internal Revenue v. Estate of Noel, 1965, 380 U.S. 678, 85 S.Ct. 1238, 14 L.Ed.2d 159, of which more will be said, we think it appropriate to set forth the considerations of fact, law, and policy which have persuaded us.

To begin, the statute which bears on this case has a reason for being, is part of a general rationale and tax law pattern, and is deliberately precise. Before the Revenue Act of 1942, the tax criterion governing cases in this area was “policies taken out” by the decedent on his own life. Section 302(g), Revenue Act of 1926, ch. 27, 44 Stat. 9. This led to difficult problems of interpretation, which the courts resolved by creating two criteria: “payment of premiums” and possession of “incidents of ownership”. The Revenue Act of 1942, ch. 619, 56 Stat. 798, Section 404, eliminated the “policies taken out" language, and sanctified the judicial gloss, with Congress, in its committee reports, including an illustrative list of the kinds of rights included under “incidents of ownership”. These included decedent's right to change beneficiaries, to borrow, to assign, to revoke an' assignment, and to surrender or cancel. H.Rep.No.2333, 77th Cong., 2d Sess., p. 164, 1942-2 Cum.Bull. 372, 491. 3

In acting this way, Congress was, we think, trying to introduce some certitude in a landscape of shifting sands. In the provision which was the predecessor of Section 2042, it was not trying to tax the extent of the interest of the decedent. That it knew how to do this is evident, for example, from a reading of Section 2033 of the Internal Revenue Code of 1954, 26 U.S.C. § 2033, which includes in the gross estate of the decedent “the value of all property * * * to the extent of the interest therein * * What it was attempting to reach in Section 2042 and some other sections was the ;power to dispose of property, the same power that the Supreme Court recognized as a basis for exercise of the tax instrument in Chase National Bank of City of New York v. United States, 1929, 278 U.S. 327, 49 S.Ct. 126, 73 L.Ed. 405. Power can be and is exercised by one possessed of less than complete legal and equitable title. The very phrase “incidents of ownership” connotes something partial, minor, or even fractional in its scope. It speaks more of possibility than of probability.

*11 Plaintiffs seize on Section 20.2042-1 (c) (2) of the Treasury Regulations on Estate Tax, which says “ * * * the term ‘incidents of ownership’ is not limited in its meaning to ownership of the policy in the technical legal sense. Generally speaking, the term has reference to the right of the insured or his estate to' the economic benefits of the policy." Plaintiffs urge that there must be “a real control over the economic benefits”. To this there are two answers. First, it is clear that the reference to ownership in the “technical legal sense” is not abandoned and supplanted by reference to “economic benefits”. Second, the regulation goes on to list illustrative powers referred to by Congress in its reports. All of these are powers which may or may not enrich decedent’s estate, but which can affect the transfer of the policy proceeds.

Viewed against this background, what power did decedent possess ? This is the relevant question — not how did he feel or act. Did he have a capacity to do something to affect the disposition of the policy if he had wanted to? Without gaining possession of the policy itself, he could have borrowed on the policy. He could have changed the method of using dividends. He could have assigned the policy. He could have revoked the assignment. Should he have gained possession of the policy by trick (as by filing an affidavit that the policy was lost), force, or chance, he could have changed the beneficiary, and made the change of record irrevocable. See Alfama v. Rose, 1949, 323 Mass. 643, 83 N.E.2d 868. Other such possibilities might be imagined. We cite these only to evidence the existence of some power in decedent to affect the disposition of the policy proceeds. In addition, he always possessed a negative power. His signature was necessary to a change in beneficiary, to a surrender for cash value, to an alteration in the policy, to a change in dividend options. Even with this most limited power, he would be exercising an incident of ownership “in conjunction with” another person. Commissioner of Internal Revenue v. Estate of Karagheusian, 2 Cir., 1956, 233 F.2d 197; Godfrey v. Smyth, 9 Cir., 1950, 180 F.2d 220; Hall v. Wheeler, D.Me., 1959, 174 F.Supp. 418.

The existence of such powers in the decedent is to be distinguished from such rights as may have existed in decedent’s father or duties owed the father by decedent. It is, therefore, no answer that decedent’s father might have proceeded against him at law or in equity. The company made it clear in the contract that it bore no responsibility for the validity of an assignment, that it could pay a beneficiary without recourse, and that it was under no obligation to see to the carrying out of any trust. It even made clear that a beneficiary need only write to the home office to receive payment. Should a third party — for example, an innocent creditor who had given valuable consideration to decedent — receive the proceeds of the policy, the proceeds of a loan on the policy, or the cash value, it could not be said that the transaction between decedent and such third person would in all such cases be nugatory. For decedent had some powers— perhaps not rights, but powers — which could, if exercised alone or in conjunction with another, affect the disposition of some or all of the proceeds of the policy.

Nor is it a compelling argument that decedent lacked physical possession of the policy. Commissioner of Internal Revenue v. Estate of Noel, 1965, 380 U.S. 678, 85 S.Ct. 1238, 14 L.Ed.2d 159; Estate of Piggott v. Commissioner of Internal Revenue, 6 Cir., 1965, 340 F.2d 829; Godfrey v. Smyth, 9 Cir., 1950, 180 F.2d 220; Hall v. Wheeler, D.Me., 1959, 174 F.Supp. 418. Moreover, as we have noted, some rights could be exercised without physical possession of the policy.

The cases arising from similar facts over nearly a quarter of a century give little support, in their holdings, to plaintiffs. Even, Estate of Doerken v. Commissioner, 1942, 46 B.T.A. 809, heavily relied upon by plaintiffs, turned on the issue whether or not decedent (as opposed to a corporation in which he had a *12 one-fourth interest) had “taken out” the policy on decedent’s life. It was not a decision that decedent possessed no incidents of ownership. Decisions in subsequent cases have, on the evidence presented, almost uniformly held the “policy facts” (reservation of rights in the policy) impregnable to attack from the “intent facts”. Estate of Piggott v. Commissioner of Internal Revenue, 6 Cir., 1965, 340 F.2d 829; Hall v. Wheeler, D.Me., 1959, 174 F.Supp. 418; Fried v. Granger, W.D.Pa., 1952, 105 F.Supp. 564, affirmed, 3 Cir., 1953, 202 F.2d 150; Estate of Collino v. Commissioner, 1956, 25 T.C. 1026; McCoy, Estate of v. Commissioner, 1961, 20 T.C.M. 224,

Estate of Morrow v. Commissioner, 1953, 19 T.C. 1068, was cited by plaintiffs as a case holding that even though the decedent designated a beneficiary who actually received one-half the proceeds of the policy, the policy proceeds were not includable in decedent’s estate. This is a misreading of the facts. In Morrow, decedent’s employer had purchased the policy and retained all rights, including the right to change beneficiaries. However, it had, as part of its employee insurance plan, undertaken to pay one-half of the proceeds to decedent’s family, and had asked decedent to designate a member or members of his immediate family. The court held that such persons were not beneficiaries under the policy, that the employer was the only beneficiary, that its contract obligation, if any, was to pay a sum equal to half the policy proceeds to such persons, and that this was not a contract of insurance and thus not within the thrust of the predecessor to Section 2042. 4

Finally, and most recently, the Supreme Court, in Commissioner of Internal Revenue v. Estate of Noel, 1965, 380 U.S. 678, 85 S.Ct. 1238, 14 L.Ed.2d 159, dealt with a factual situation at least as favorable to the decedent’s estate as that in this case. In Noel, two airline flight insurance policies were purchased on decedent’s life, the two premiums of $2.50 each being supplied by decedent’s wife. Decedent was quoted as having instructed the sales clerk to “give them to my wife. They are hers * * * I no longer have anything to do with them.” The contracts reserved to the decedent the rights of assignment and change of beneficiary. Several hours later decedent died when his plane crashed. Three alternative arguments were made to support the conclusion that decedent had reserved no incident of ownership in the policies: (1) that decedent’s wife purchased the policies and therefore owned them; (2) that, even if decedent had owned them, he had given them to her; and (3) that even if he had reserved contractual powers, such powers were illusory, being impossible to exercise between take-off and crash.

The Court rejected the first contention by saying that the policies themselves rebut the wife’s claim of complete ownership with an irrevocable right to remain the beneficiary. As to the second, it held that the power to assign the policies remained in the decedent at the time of his death since the alleged assignment had not been endorsed on the policies as was required.

The third point, that of illusory power to exercise the retained legal rights, was answered in the following language, 380 U.S. at 684, 85 S.Ct. at 1241:

“We hold that estate tax liability for policies ‘with respect to which the decedent possessed at his death any of the incidents of ownership’ depends on a general, legal power to exercise ownership, without regard to the owner’s ability to exercise it at a particular moment.”

*13 To the principle of heavy predominance of the “policy facts” over the “intent facts” there must be added the caveat that, where the insurance contract itself does not reflect the instructions of the parties, as where an agent, on his own initiative, inserts a reservation of right to change a beneficiary contrary to the intentions which had been expressed to him, no incidents of ownership are thereby created. National Metropolitan Bank of Washington v. United States, 1950, 87 F.Supp. 773, 115 Ct.Cl. 396; Schongalla v. Hickey, 2 Cir. 1945, 149 F.2d 687. The case before us presents no such issue, for the right in decedent to change beneficiaries was recognized on the one occasion when it was exercised and this right continued thereafter.

While decisions against the estate of a passive but power-possessing decedent may often conflict with the honest intentions and understanding of premium-paying beneficiaries and insureds, the alternative of abandoning the insistence on the governing nature of the contract, in most cases, is less desirable. The drawing of a useful line would be impossible; there would be a much wider range of varying decisions on similar facts; and there would be an invitation to unprincipled estate manipulation. As government counsel has pointed out, there could always be a formally executed side agreement under which the insured clearly surrenders to the beneficiary all his rights to the policy, such agreement to be brought to light only in the event of the decedent’s dying before the beneficiary.

In any event, the statute has been on the books since the Revenue Act of 1942. This is only one of a number of cases applying it in the face of considerable external evidence of intent. Charles Horton, who caused the policy to be taken out, saw fit to vest decedent with rights in the policy and to allow such rights to continue for thirty-four years. Charles was a successful businessman and with as much incentive, opportunity, and capacity to be aware of the laws of the land as most people. It is difficult to speculate what purpose he thought was being served by his son’s retention of rights in the policy. Had he wished to deprive his son of all incidents of ownership in the policy, this result could easily have been accomplished. But the step was not taken. We find that the decedent died, possessing at least an incident of ownership in the policy on his life.

Judgment will be entered vacating the judgment of the district court and ordering judgment for the defendant.

. The relevant part of the taxing statute, Section 2042, is as follows:

“The value of the gross estáte shall include the value of all property—

He $ $ H*

“ (2) Receivable by other beneficiaries. —To the extent of the amount receivable by all other beneficiaries as insurance under policies on the life of the decedent with respect to which the decedent possessed at his death any of the incidents of ownership, exercisable either alone or in conjunction with any other person. >k ❖ »

. It is nóted thát with regard to the unrestricted option (Option A), the policy application contains the following words:

“If a [sic] is chosen the policy may become part of the estate in case of bankruptcy”.

. Subsequently, the Revenue Act of 1954 eliminated the premium payment test, leaving possession of “any incidents of ownership” as the sole criterion. 26 U.S. C. § 2042. Sen.Rep. No. 1622, 83d Cong., 2d Sess., p. 472 et seq., U.S.Code Cong. & Admin.News 1954, p. 4629.

. The reservation of a right in the insured to change the beneficiary was found to be of no significance and the policy proceeds were held not includable in his estate in Lamade v. Brownell, M.D., Pa., 1965, 245 F.Supp. 691. However, in that case, unlike the one before us, there existed an absolute assignment of the policy, the assignees of the policy were recognized by the insurer as the owners of the policy and could act independently of the insured, and had the insured attempted to change the beneficiary such attempt would have been ignored by the insurer.