The Social Security Administration ("SSA") had to decide whether Amy Marcoux was the legitimate daughter of Gene Marcoux or the illegitimate daughter of the late Raymond Petit. If the latter, the SSA will pay Amy a share of Raymond Petit's survivor's benefits, although in order to do so, it will have to reduce the current payments it makes to Raymond's two legitimate children. See 42 U.S.C. Secs. 402(d), 403(a). These two children, through their mother (Barbara Becker), have opposed Amy's benefit application. See 20 C.F.R. Sec. 404.932 (1989). After hearing testimony from Amy's mother, Lisa Marcoux, from Lisa's mother, from Raymond's sisters, and from other relatives and acquaintances, an Administrative Law Judge decided that Amy was illegitimate, that Raymond Petit was Amy's father, and that Amy should receive benefits.

The Appeals Council of the Department of Health and Human Services reviewed the ALJ's findings and sent the case back to the ALJ. A federal statute requires the HHS, in

determining whether an applicant is the child or parent of a[n] ... insured individual, ... [to] apply such law as would be applied in determining the devolution of intestate personal property by the courts of the State in which such insured individual ... was domiciled ... at the time of his death.

49 U.S.C. Sec. 416(h)(2)(A). Under New Hampshire law, (as the Appeals Council read it),

a child born to married parents is presumed to be legitimate. This presumption can be rebutted by clear and convincing evidence. However, this evidence may not consist of statements or testimony from the child's mother or her husband.

Fearing that the ALJ had improperly relied on testimony given by Amy's mother Lisa, the Appeals Council remanded for further consideration.

On remand, the ALJ disregarded both Lisa's testimony and a letter written by Eugene Marcoux. It then reached the same conclusion as it had before. Relying solely on record evidence provided by others, the ALJ found clear and convincing evidence that Eugene Marcoux was not Lisa's father, and that Raymond Petit was. This time the Appeals Council affirmed. Raymond's other children then asked the federal district court to review the federal agency's findings. The court did so. It concluded that the evidence before the agency was not strong enough to show that Amy was Raymond's child, and it set aside the agency's determination. Amy's mother (Lisa), acting on Amy's behalf, now appeals that decision to us.

We shall assume, for the sake of argument, that the statute we have quoted above, 49 U.S.C. Sec. 416(h)(2)(A), requires HHS to apply, not only New Hampshire's substantive law, including presumptions and burden-of-proof rules, but also special evidentiary rules such as the one that limits a mother's testimony on the question of her own child's legitimacy. Specifically, we shall assume:

(1) that, because Lisa Marcoux was married to Gene Marcoux when Amy was born, the agency must presume that Amy is legitimate. See Twomey v. Twomey, 116 N.H. 29, 31, 351 A.2d 66 (1976);

(2) that to overcome the presumption, Amy must establish, by clear and convincing evidence, that Raymond Petit was her father. See N.H.Rev.Stat.Ann. Sec. 561:4;

(3) that, in doing so, Amy cannot use her mother's testimony to help rebut the presumption. We recognize that the extent to which this evidentiary rule, known as Lord Mansfield's Rule, bars a mother's testimony is open to argument. Does it bar all testimony that casts doubt on legitimacy or only testimony about a husband's "access"? See State v. Sargent, 100 N.H. 29, 31, 118 A.2d 596 (1955); Saunders v. Fredette, 84 N.H. 414, 418, 151 A. 820, 824 (1930); Parker v. Way, 15 N.H. 45, 49 (1844); see generally Michael H. v. Gerald D., --- U.S. ----, 109 S. Ct. 2333, 2342-43, 105 L. Ed. 2d 91 (1989). We need not answer this question, for the Appeals Council found that the ALJ, the second time around, based his conclusions, not on Lisa's testimony, but on other evidence in the record, such as "information provided by the families and friends of Lisa Marcoux and Raymond Petit." It also found that this other evidence, taken by itself, constituted "clear and convincing evidence" that Eugene Marcoux was not Amy's father. In our view, the record adequately supports this final HHS conclusion.

The precise legal question before us is whether, on the basis of the whole record, we can say that "substantial evidence" supports the agency's factual conclusion. See 49 U.S.C. Sec. 405(g) ("The findings of the Secretary as to any fact, if supported by substantial evidence, shall be conclusive"); Richardson v. Perales, 402 U.S. 389, 401, 91 S. Ct. 1420, 1427, 28 L. Ed. 2d 842 (1971) ("substantial evidence" means " 'such relevant evidence as a reasonable mind might accept as adequate to support a conclusion' " (quoting Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S. Ct. 206, 217, 83 L. Ed. 126 (1938))); Universal Camera v. NLRB, 340 U.S. 474, 477-91, 71 S. Ct. 456, 451-66, 95 L. Ed. 456 (1951) (holding that "substantial evidence" review requires consideration of the record as a whole); Allen v. Califano, 613 F.2d 139, 145 (6th Cir.1980) (same). In light of New Hampshire law, the question looks awkward: "Is there substantial evidence that would permit the agency to find by clear and convincing evidence that Amy was Raymond's daughter?" But to apply this legal standard is not particularly awkward as long as one remembers its point, namely, that Congress has entrusted the agency, not the court, with the factfinding job. A reviewing court must treat the agency's factual conclusion with considerable respect, indeed more respect than an appellate court would show a factual conclusion reached by a district judge, for the agency has not only had an opportunity to view the witnesses and determine their credibility, but it also has special expertise, experience, and knowledge of the subject matter that guide its determination of the facts. See Consolo v. Federal Maritime Comm., 383 U.S. 607, 620, 86 S. Ct. 1018, 1026-27, 16 L. Ed. 2d 131 (1966) ("substantial evidence" review "gives proper respect to the expertise of the administrative tribunal"); Penasquitos Village, Inc. v. NLRB, 565 F.2d 1074, 1078-79 (9th Cir.1977) ("Weight is given the administrative law judge's determinations of credibility for the obvious reason that he or she 'sees the witnesses and hears them testify, while the ... reviewing court look[s] only at [the] cold record[ ].' " (quoting NLRB v. Walton Mfg. Co., 369 U.S. 404, 408, 82 S. Ct. 853, 855, 7 L. Ed. 2d 829 (1962))); Orvis v. Higgins, 180 F.2d 537, 540 (2d Cir.1950) ("evidence sufficient to support a jury verdict or an administrative finding may not suffice to support a judge's finding"); K. Davis, 5 Administrative Law Treatise Sec. 29:26, at 455 (2d ed.1984). Thus, the real question for a court is whether a court, keeping these considerations in mind, could say that the agency could not find, by "clear and convincing evidence," that Raymond was Amy's father. After examining the record, we conclude that the district court was wrong to set the agency's determination aside.

The record contains highly detailed testimony from Lisa's mother, Lisa's sister, and Raymond's sisters. Some, or all, of them testified in detail about Lisa's relationships with her husband, Eugene, and with Raymond Petit. They said that Lisa had separated from Eugene in 1982, that Eugene did not live in New Hampshire, that Lisa had lived with Raymond between 1982 and June 1984, that she was living with Raymond in March 1984 when Amy was conceived, that Raymond had told them that he and Lisa wanted a child, that Raymond had told them that he and Lisa had conceived a child that Lisa miscarried, that he told them Lisa was again pregnant with Amy, that he began to live separately from Lisa in June 1984 (because Lisa and he fought about his drinking problem), that he nonetheless provided Lisa with money (when available) to help support Amy, as well as with diapers and baby bottles, that he was proud of Amy, that he said he was Amy's father, that he brought Amy Christmas and Easter presents, and that he visited Amy. This testimony is sufficiently detailed, and those who provided it seem sufficiently aware of the circumstances, to permit a factfinder who believed it to conclude that Raymond was Amy's father. And the ALJ, who heard the testimony, had every right to believe it. See Rodriguez v. Secretary of Health And Human Services, 647 F.2d 218, 222 (1st Cir.1981) (" 'Issues of credibility ... are the prime responsibility of the Secretary' " (quoting Rodriguez v. Celebrezze, 349 F.2d 494, 496 (1st Cir.1965))).

The record also contains contrary evidence, but none of that evidence compels a contrary finding. First, in 1979 Raymond had a vasectomy. The ALJ was free to disregard this fact, however, for a Dr. Conway, in 1982, performed a vasectomy reversal. Dr. Conway wrote that he could not say for certain whether Raymond's particular reversal worked, because Raymond never returned for a "post-operative sperm count;" but he did say that the "vasectomy reversals that we have performed have generally been successful...."

Second, Amy's birth certificate says that Eugene Marcoux is her father. The ALJ need not have given weight to this fact, however, for Lisa's mother testified that "the hospital has the policy that if you're married you have to have a person your married to's name. You can't have a baby under another name;" and Lisa's sisters also testified that (whatever the hospital's actual policy) Lisa thought she had to give Eugene's name irrespective of the actual father.

Third, Barbara Becker, Raymond's wife and the mother of the legitimate children, testified that Eugene was in New Hampshire in April 1984, just after the time Amy was conceived, that he visited Lisa, and that Lisa and Raymond's relationship, which had been sporadic, was stormy at that time. Of course, the ALJ was free to disregard Barbara Becker's testimony completely. See Rodriguez, supra. Moreover, testimony from Lisa's sister and mother weakened the impact of Barbara's testimony. They explained that Eugene and Lisa had two children before Amy was born, that Eugene occasionally would "come back on leave" from the Navy to visit these children, that "usually [their] bags were already packed, and they were all ready to go when he got there," and that he "used to take them over to his brother's house, pick them up and go right over there and spend the weekend."

Fourth, the record contains some other, miscellaneous evidence, such as that Eugene's insurance paid for Amy's birth, and that various individuals to whom Raymond had spoken about his children said that he did not mention Amy. This evidence is weak, barely significant, on the paternity question.

In sum, having reviewed the record on the assumptions that the Appeals Council made--that Lisa's testimony would not be counted--we conclude that it legally supports the agency's decision. Given the broad scope of the agency's legal power to make factual determinations, we must uphold, on the basis of the testimony and related evidence (Lisa's aside), its conclusion that the evidence clearly and convincingly shows Raymond is Amy's father.

Reversed.