SIMONS, Circuit Judge.

The appeal is by an insurer from a judgment on verdict of the jury in favor of the beneficiary under two policies of life insurance, for double indemnity for accidental death, and for permanent and total disability benefits, the insurer having paid the ordinary life benefits prior to suit. The principal error assigned is the failure of the court below to grant the insurer’s motion for peremptory instructions.

The insured, James R. Yaughan, died on the 19th day of November, 1929, the immediate cause of death being a pulmonary em-bolus. The appellee, his widow, is the beneficiary named in the policies. Yaughan was a contractor, and on July 19, 1929, while inspecting work on which he was engaged, 'fell into- a ditch and fractured his left leg. On August 25, 1929; while still using crutches as the result of the fracture, he slipped and fell in the bathroom of his home. Shortly thereafter he began to have pains in his chest, and was removed to a hospital, where he had several hemorrhages of the lung, and where, eighty-six days after his second fall, he died. An autopsy revealed thrombosis in the left leg, from which portions had broken off and had circulated in the blood stream. One of *395these portions, or emboli, was found to occlude the pulmonary artery, causing the death.

The policies called for the payment of double indemnity, provided:

“1. That the death of the insured shall result directly, exclusively and independently of all other causes, from bodily injury effected solely through external, violent, and accidental means, and shall not be caused or contributed to directly or indirectly, wholly or partially, by disease; and

“2. That death shall occur within ninety days from the occurrence of such injury. ■s- * *»

If the first aeeident was the cause of death, or a contributing cause, there is no liability, because death did not result within ninety days from the injury caused by the first fall. The beneficiary’s theory, however, is that the thrombus which had developed in the insured’s leg from the fracture had become absorbed or organized before the bathroom fall; that the latter caused the formation of new clots, and initiated the detachment of the em-bolus whieh obstructed the pulmonary artery and caused death. She contends there is substantial evidence to support her view, and that the fact issue is concluded by the verdict of the jury. If she is wrong in this, and if the thrombus resulting from the leg fracture on July 19, 1929, persisted to the .dato of the second aeeident, then it is her contention that as a matter of law the thrombus was merely a condition and not in any sense a cause of death, the sole and proximate cause thereof being the accidental fall in the bathroom on August 25, 1929.

Five physicians testified as to the cause of the insured’s death. Of these, Dr. Sanders and Dr. Keller were in attendance prior to death, Dr. Woodruff conducted the autopsy, and testified from the facts by it disclosed, and Dr. Goodpasture and Dr. Tigert qualified as experts. All of the medical testimony, except that' of Dr. Sanders on his direct examination, was in aeeord to the effect that the leg fracture was a contributory cause of death. Dr. Keller signed the death certificate on November 20,1929, after being present at the autopsy on the previous day. He certified that the fracture of the leg was a contributory or secondary cause of death, and reaffirmed this conclusion as a witness. Dr. Woodruff was of the opinion that the first thrombosis must be attributed to the direct injury to the smaller veins of the ankle at the time of the fracture, as well as to inactivity of the left leg following the first accident. The fall in the bathroom did not in his opinion dislodge the embolus whieh brought about the death. The time interval was too great to have dislodged the large embolus which he found at the autopsy. In addition to the medical testimony, Haley Vaughan, son of the insured, the first one to see him after his fall in the bathroom, testified that his father had said that in falling he had struck Ms hip.

In his direct examination Dr. Sanders testified that in Ms opinion the embolus which occluded the pulmonary artery and caused the insured’s death was caused or in some way produced by an injury to the left leg a.t the time of the bathroom fall, and that the earlier accident which produced the fracture did not in any respect contribute to it. Standing alone, and upon a factual rather than a conjectural basis, -this evidence would have supported the verdict. - On cross-examination, however, Dr. Sanders testified that it is impossible to look at the outside of a man’s leg and tell whether it is in a thrombosed condition if tlie smaller veins are involved; that he could not sa.y whether the thrombus inside the bone was dissolved or whether it con tinued after the insured fell into the ditch, and that nobody could tell. Ho testified that it was entirely possible for an embolus from the thrombus which followed the fracture to have traversed the blood stream into- the pulmonary artery, and that it was entirely possible for the death to have had that particular origin. When pressed for the basis of his original opinion, he explained that he first assumed there was a blow on the left leg somewhere, though ho didn’t look for it. He next assumed that an embolus had broken from a thrombus in the left leg. Ffom this assumption he had to conclude that the thrombus was caused by some disease o-r blow. Assuming the insured had no diseased condition, a further conclusion was that there was a traumatic injury to the blood vessel as a result of the second accident, though he found no evidence of that fact.

We conclude from this that Dr. Sanders’ evidence on cross-examination was completely destructive of Ms original opinion, and, there being no other proof, there was no substantial evidence to submit to the jury on the theory that the second accident was the sole cause of death. While the courts will give wide latitude to- the reception of expert opinion evidence, we think it axiomatic that if: must be based upon conceded or proved facts, and that a naked opinion, based obviously on mere speculation and conjecture, does no-t rise to the dignity of evidence, espe*396cially when it is in conflict with the conceded physical facts disclosed by the autopsy.

We come then to the contention that the thrombosis which followed the fracture was a mere condition during the existence of which death occurred, but which did not cause it; the bathroom fall being the sole, moving, and proximate cause of the dislodging of the em-bolus and the resulting death, and reliance is 'placed upon Manufacturers’ Accident Indemnity Company v. Dorgan, 58 F. 945, 2.2 L. R. A. 620 (C. C. A. 6), and Mutual life Insurance Co. v. Dodge, 11 F.(2d) 486, 591 A. L. R. 12901 (C. C. A. 4). The first of these cases involved a situation wherein a temporarily defective heart action caused a fall into the water, where drowning resulted, and the second is the well-known novocaine case [see Pope v. Prudential Insurance Co., 291 F.(2d) 185 (C. C. A. 6), and Maryland Casualty Company v. Massey, 38 F.(2d) 724, 71 A. L. R. 1428 (C. C. A. 6)], whereinhypersuseeptibility to the drug rendered a normal application of it fatal. Neither the disease in the one ease nor the personal idiosyncrasy in the other was recognized as a cause of death. Neither of the cases is apposite. If in the one case the deceased suffered death by drowning, the drowning was its proximate cause, no matter what caused the fall into the water, and even though one is susceptible to novocaine, it is the administration of the drug that causes the death none the less.

The test here is whether the death is directly traceable to the first injury and attributable to it as an active, direct, and contributing cause of it. Had Vaughan’s fall in the bathroom fractured his skull or broken his neck, and as a result he had died, the Dorgan and Dodge Cases might have been controlling, for in that event there would have been no death-causing injury attributable to the first fall, however its results may have placed him in a position where he met with the second accident. Another consideration contributes to our conclusion. Had Vaughan’s second accident occurred and death resulted within ninety days of his fall into the ditch, we have no doubt that liability could have been based and recovery had ■upon the theory that the first fall was the accidental cause of death. If we are right in this, it follows that the first fall was a contributing cause, even though, unfortunately for the beneficiary, it falls without the terms of the policies. We think the District Judge was in error in denying the motion for directed verdict.

There is also in the case a question of the timeliness of the notice to the insurer, and a claim of waiver by the beneficiary. In view of our conclusions on the merits, we find it unnecessary to give consideration to them.

The judgment below included the sum of $300 for disability benefits on the theory that the insured was totally'disabled from the time of the bathroom fall to the date of death. No proof of such disability was furnished the insurer, as required by the policies. This is a condition precedent to recovery, and Bergholm v. Peoria Life Insurance Company, 284 U. S. 489, 52 S. Ct, 230, 76 L. Ed. 416, is controlling on that issue. The judgment in this respect must also be set aside.

Reversed and remanded for new trial.