This is an appeal by the self insurer in a workmen’s compensation case from a decree awarding compensation to the dependents of one Sevigny. The single member found that there existed “a probable causal relationship between the employee’s death and his industrial injury” of November 21, 1955, which finding was in effect[*748] adopted by the reviewing board. The decision of the board also considered motions by the self insurer to recommit and to correct the transcript. Other than correcting the Christian name of the self insurer’s expert, both motions were denied.

The employee, Royal E. Sevigny, while employed as a cook by Harvard University, received a cut on his right index finger when a knife which he was using to slice meat slipped. He received treatment the same day at the Harvard University Employees’ Clinic and additional treatment on November 23, 25, 28, 29 and 30, and December 1, 1955. On the last mentioned date he was admitted to the Faulkner Hospital and was treated for a staphylococcus aureus infection of the second metacarpophalangeal joint of his right hand. The wound was incised and drained and he was discharged on December 9, 1955. He thereafter was given physiotherapy for his finger and it was examined at the clinic on several dates through January 17, 1956. There was oral testimony that about a month after his return from his first visit to the hospital, the employee, who had always been active about the house, had become listless and complained of being tired. His gums were bleeding and his palate was large and swollen. On Sunday, January 22,1956, he remained in bed, too tired to accompany his wife to church as he usually did. The next day she saw that his face was swollen and he had a cough and a temperature. He was then seen by the family physician on January 23, 1956, and was readmitted to the hospital on the same day where a diagnosis of monoblastic leukemia was made. He died on January 25, 1956.

The question presented by this appeal is whether or hot there was sufficient proof of a causal relation between the contracting of monoblastic leukemia and the preceding injury. The case seems not to have been heard on any issue of acceleration or aggravation of leukemia as a preexisting disease. Cf. Ralph’s Case, 331 Mass. 86, 89; Murphy’s Case, 328 Mass. 301.

The findings and decision of the single member as affirmed[*749] by the reviewing board are to be sustained wherever possible unless they are wholly lacking in evidential support or contain an error of law. Lysaght’s Case, 331 Mass. 451, 452-453. Br ek’s Case, 335 Mass. 144, 147. See also Hartman’s Case, 336 Mass. 508, 511, and cases cited. The burden is upon the dependents to prove that the staphylococcus infection was the probable cause of the employee’s leukemic condition, see Sulham’s Case, ante, 586, 589, although the evidence need not exclude all other possibilities. Blanchard’s Case, 277 Mass. 413, 415. And where such causal relation is a matter beyond the common knowledge and experience of the ordinary layman, proof of causation between the injury and the ensuing death must rest upon expert medical testimony. Josi’s Case, 324 Mass. 415, 417-418. Murphy’s Case, 328 Mass. 301, 303. Ralph’s Case, 331 Mass. 86, 90. Charron’s Case, 331 Mass. 519, 521. LeBlanc’s Case, 334 Mass. 265, 267. See Lovely’s Case, 336 Mass. 512.

As the findings of the single member and the board are based on the expert testimony of Dr. Stefanini, who was called as a witness by the dependents, it is necessary to review his testimony. This witness testified that acute leukemia is a form of malignant cancer with wild growth which affects the bone marrow and results in an increase in the white blood cells and a reduction of normal red blood cells so that a patient develops anemia. He expressed his opinion that the employee was suffering from monoblastic leukemia which is known for a very rapid course so that a patient survives for a shorter time after attack than with other types of leukemia. When asked his opinion as to the causal connection of the injury and the ensuing infection with the resultant death by leukemia he replied that “there is a possible causal relationship between the infection and the development of acute leukemia in this case.” If the evidence stopped at this point the opinion would be no more than mere speculation or conjecture and would have to be disregarded. Falco’s Case, 260 Mass. 74, 77. Green’s Case, 266 Mass. 355, 357. Perangelo’s Case, 277 Mass. 59, 65. Johnson’s Case, 278 Mass. 365, 369. Halnan v. New England [*750] Tel. & Tel. Co. 296 Mass. 219, 223. Cf. Ralph’s Case, 331 Mass. 86, 90. See also Atlas v. Silsbury-Gamble Motors Co. 278 Mass. 279, 282-283. Evidence must be sufficient to create actual belief in the fact to be proved, Tartas’s Case, 328 Mass. 585, 587, and evidence that tends equally to support two inconsistent propositions, one of liability and the other of nonliability, will not support either proposition. See Sargent v. Massachusetts Acc. Co. 307 Mass. 246, 251. Thus the evidence must be such that it was more likely that there was a causal relationship between the death and the injury. Otherwise the evidence must be disregarded as conjectural. See Atlas v. Silsbury-Gamble Motors Co. 278 Mass. 279, 282-283; Tartas’s Case, 328 Mass. 585, 587, and cases cited.

The dependents’ expert finished his direct examination by answering a hypothetical question whether the death of the employee was a probable result of the injury on November 21, 1955, after being informed that “probable” in the question meant “more than possible.” The witness answered, “It was a probable outcome.” On cross-examinatian the witness was asked whether it was not a matter of speculation to state that the infection could have caused the leukemia. To this question the witness replied that they have a man (the deceased) who is well up to November 21, 1955, and a blood count is taken somewhere between December 1 and December 9, and he is still perfectly well as far as his blood is concerned, then in a matter of six weeks “he explodes in a disease” which they know can develop very suddenly. When asked for a yes or no answer, the witness replied that he did not think he could answer, but if counsel wanted a direct answer “he will say yes” ^[1] (i.e., it is not a matter of speculation). This testimony, highly ambiguous, together with a previous answer that the leukemia was more than a possible result of the infection, could have been[*751] regarded by the board as testimony to the effect that the witness believed that the death by leukemia was more likely than not a result of the staphylococcus infection. The fact that the witness further testified that “It is not unlikely [that the infection would have been the cause of the onslaught of acute leukemia in Mr. Sevigny’s case]]” confirms this interpretation. DeFilippo's Case, 284 Mass. 531, 534-535. Josi’s Case, 324 Mass. 415, 418-419.

However, it is contended by the self insurer that since the state of medical knowledge as to the causes of leukemia is indefinite, any opinion based on such medical knowledge is of necessity merely speculative and conjectural. See Hogan v. Stovall Drilling Co. Inc. 55 So. 2d 284 (La. App.); Kundiger v. Waldorf Paper Prod. Co. 218 Minn. 168; Beatrice Creamery Co. v. Kizer, 127 Neb. 34; Boyer v. Department of Labor & Indus. 160 Wash. 557. The courts are not to determine which side of a medical dispute is sound where each side is supported by reason and logic (see Duggan’s Case, 315 Mass. 355, 358; Murphy’s Case, 328 Mass. 301, 304), but an opinion given by an expert will be disregarded where it amounts to no more than mere speculation or a guess from subordinate facts that do not give adequate support to the conclusion reached. Ruschetti’s Case, 299 Mass. 426, 431. Callaghan v. R. H. White Co. 303 Mass. 413, 416. Brownhill v. Kivlin, 317 Mass. 168, 170. Nass v. Duxbury, 327 Mass. 396, 401. Ralph’s Case, 331 Mass. 86, 90. Gladstone v. Treasurer & Receiver General, ante, 48, 51. In those cases ^[1] the opinions did not have probative value because the basic facts upon which they were based were either gratuitously assumed (see Brown v. United States Fid. & Guar. Co. 336 Mass. 609) or were otherwise insufficient to form the basis of a reasoned opinion. See Model Code of Evidence, Rule 303, comment. But even though an expert opinion be unequivocal and even though it is based on substantial data, the conclusion may still be conjectural and speculative where it is contrary to common[*752] knowledge or common sense. Pastrick v. S. S. Kresge Co. 288 Mass. 194. Callaghan v. R. H. White Co. 303 Mass. 413, 416. Lavoie v. Brockelman Bros. Inc. 315 Mass. 673. Gladstone v. Treasurer & Receiver General, ante, 48, 51. See also Ruschetti’s Case, 299 Mass. 426, 431. Cf. Friese v. Boston Consol. Gas Co. 324 Mass. 623, 628; Brett v. J. M. Carras, Inc. 203 F. 2d 451, 453-454; Deitz v. United States, 228 F. 2d 494, 495; Wigmore, Evidence (3d ed.) § 662. The instant case is more like our cases last cited in that the expert’s underlying knowledge is questioned because of the lack of medical knowledge. See Blanchard’s Case, 277 Mass. 413, 415.

. During the examination of Dr. Stefanini frequent references were made to treatises written by medical authorities dealing with the relationship between bacterial infection of the type admittedly incurred by the employee and acute leukemia of the kind which caused his death. Copious abstracts of these publications were attached to the motion to recommit which was presented and considered by the board and which in reality was a motion for a new trial. These abstracts were expressly made a part of the record by order of the board. No question is raised as to their accuracy. See Bacon v. George, 216 Mass. 519, 520; Davis v. Boston Elevated Ry. 235 Mass. 482, 495; Samuel v. Page-Storms Drop Forge Co. 243 Mass. 133, 135. One authority to whom Dr. Stefanini referred in his testimony as the author of a book which is the envy in the field (the relation of infection to leukemia) because it “pretty much is the bible of all in that field” states, “The cause of leukemia is, thus, unknown and the pathogenesis of the disorder is obscure.” A second American author who believes the cause of cancer is unknown does not agree with the holding of another author that in the case of children there is a relation between infection and the disease. The employee was forty-three years of age. Finally, reference was made to the work of two French clinicians who state that acute leukemia “may follow” severe infection. Doctor Stefanini himself as a witness said that “there is an emphasis on the ‘may.’ ” It did not appear[*753] that the opinion of the dependents’ expert had too secure a footing from these quoted authors.

The expert for the dependents testified that he had seen five hundred cases of leukemia and that in six of them an “infection was a possible producing cause” (emphasis supplied). He also stated that the opinions of medical authorities are about equally divided whether there is a causal connection between infection and leukemia (“50-50 all along the line”). He apparently recognized that there was less evidence that bacterial infection might be a contributing cause of leukemia than that viruses might be such a cause. ^[1] The expert relied considerably on medical writings, set forth in the record, to support his views. These writings, however, indicate that the matter is still an unresolved medical problem. In the expert’s own experience in about one per cent of the cases observed by him, there was a possibility, but no more than that, that bacterial infection was a contributing cause.

The evidence taken at its best for the dependents does not go beyond showing a possible cause. It comes down to this: There is some basis for the hypothesis that staphylococcus infection is a cause of leukemia, but so far it is only an unproved hypothesis, which the dependents’ expert thinks may sometime be proved. This is not proof of cause. Cases are to be distinguished where the expert opinion is that a medically accepted cause was operative to relate the accident to the injury. Sheppard’s Case, 287 Mass. 459, 463. Robinson’s Case, 299 Mass. 499, 502. Geagan’s Case, 301 Mass. 319. O’Connor v. Griff, 307 Mass. 120, 123. Duggan’s Case, 315 Mass. 355, 358. Comeau v. Beck, 319 Mass. 17, 19-20. See also Black v. Boston Consol. Gas Co. 325 Mass. 505, 508.

We think that upon the entire evidence the dependents[*754] have failed to sustain the burden of proof in attempting to prove the cause of the disease, where the cause is admittedly unknown, by proof of a particular cause. It follows that the decree must be reversed. Falco’s Case, 260 Mass. 74. Green’s Case, 266 Mass. 355. Perangelo’s Case, 277 Mass. 59. Johnson’s Case, 278 Mass. 365. Halnan v. New England Tel. & Tel. Co. 296 Mass. 219. Ruschetti’s Case, 299 Mass. 426. Ralph’s Case, 331 Mass. 86. LeBlanc’s Case, 334 Mass. 265.

bo ordered.