The deceased was a man 69 years of age weighing about 200 pounds and with an aggravated heart condition of long standing who had been hospitalized while receiving treatment for heart and respiratory ailments, *906to which treatment he was responding normally. On the night in question he was alone in the hospital room on a bed 34 to 36 inches high protected on each side by guard rails rising 14 to 18 inches above the mattress. He was checked every hour and found to be sleeping normally until about 4:30 a.m., a half hour after the last observation, at which time a nurse found him lying face down on the floor, dead, near the foot of the bed, with his feet toward and his head away from the bathroom and entrance doorways, arms folded under him. The sock on his left foot was torn; the toenail of the big toe of the left foot was split and torn from the toe, the toe of the sock was drenched in blood and the nurse who found him estimated there was a half of a measuring cup of blood on the floor. There were no other marks on the body, no indications of violence on the bed, and nothing in the bed which suggested the manner in which the toe could be so injured that the nail was torn from the flesh.
A medical witness testified in part: “Assuming that at that time the big toe on his left foot was bleeding around the nail and blood was on his stocking and also on the floor, in my opinion injury of some type would cause that bleeding. . . Assuming that he was in that position and those facts had occurred, and also knowing his physical condition with arteriosclerosis, congestive heart failure and his previous history of heart trouble, assuming that Mr. Halligan fell, that he fell heavily to the floor of his room and injured his toe to the extent that it bled, as to what effect that would have on his body and on his physical constitution, I think it might well have killed him. . . We know this, that any movement which would involve a great bodily stress, such as trying to catch oneself on his arms as he fell forward, indeed even the excitement and the sudden shock of falling, might well be injurious to the individual. . . I must say that, if a man weighing close to two hundred pounds fell on the floor, while it could have no effects, it might certainly lead to fatal results. . . Assuming that he died of acute heart failure, in my opinion the fall to the floor certainly could have caused acute heart failure. In my opinion as a physician, under those circumstances, as to whether or not *907it would have been the most likely cause, I think that’s what killed Mr. Halligan.” (Emphasis added.) He also testified: “I testified this morning that Mr. Halligan had recovered from one of these coronary attacks, which left him with this infarction, several years prior to his death. Patients recover from such attacks far more frequently than they die from them. . . Mr. Halligan had a coronary thrombosis. . . It’s my firm belief that if you took one hundred people with a coronary occlusion and threw them heavily on the floor . . . there would be a fifty percent increase in mortality, Now, if an individual had an acute coronary thrombosis, and acute occlusion of the coronary arteries, and if it were of sufficient intensity to kill him, and if it did kill him, that would immediately stop the blood from circulating. . . As to whether a wound on such a person’s toe would bleed after such an attack which caused the heart to stop beating immediately, he might have oozing of a drop or two of blood, but he would not bleed in the ordinary sense of the word.”
■ From all of the evidence introduced, a finding is demanded that at some time between four and four thirty or five a.m. the patient got out of bed without the guard rails being let down. There was testimony that it is much more likely for a person who suffers the onset of a heart attack to immediately sit or lie down than for him to stand up or walk about; therefore it is a logical inference that the heart attack developed after, rather than before, the patient commenced to get out of bed. There is undisputed evidence that he fell to the floor (though whether he fell from the bed or only from a standing position is not shown) and there is undisputed evidence that he was wounded by catching his toe in something with sufficient force to tear the sock and split and rip the toenail so severely that he lost a half of a measuring cup of blood. Nothing in the bed suggested itself as causing the injury. There is undisputed evidence that the deceased, after getting out of the bed, remained alive long enough to bleed a half cup of blood from a small wound in the big toe. There is medical opinion evidence that the patient’s fall was the cause of his death, considering his weight, age, and physical condition.
*908“Only when the plaintiff’s evidence does fail to prove the case as laid in the petition, without revealing as defense matter fatal to the cause pleaded, or where the evidence adduced by the defendant, as a matter of law, conclusively refutes the proof made of the plaintiff’s case, can a verdict for the defendant be directed.” Moate v. H. L. Green Co., 95 Ga. App. 493, 500 (2) (98 S. E. 2d 185). It is strenuously contended that the plaintiff’s evidence points equally toward opposing theories; that is, whether a fall caused the heart attack or whether a heart attack caused the fall. If the latter, of course, the death would not be the result of accident within the meaning of the policy, but if the former, the plaintiff could recover under the terms of the policy, which does not demand either that an accident cause the death independently of all other causes, or that the accident be evidenced by violent, visible and external means. The direct opinion evidence of one medical witness was that a fall to the floor caused acute heart failure. It is argued, however, that this witness must assume a fall before inferring it as the cause of the heart attack (the heart attack being admittedly the immediate cause of death) and that this is pyramiding an inference on an inference. In Lumbermen’s Mutual Casualty Co. v. Bridges, 81 Ga. App. 395, 401 (58 S. E. 2d 849) it was stated in a similar state of facts: “We think that these inferences are collateral, virtually only one inference, and that the problem of basing one inference on another is not involved. But if it is true that the inference that the exertion contributed to the death is an inference based on the inference that the employee died of heart disease, neither inference is too remote and complies with the test of validity, having as a basis the connection of cause and effect and the observations of human experience.” The witness here, in arriving at his opinion, would necessarily weigh certain probabilities, as also would the jury in reaching a verdict in the case. The probabilities, from the witness’s point of view, concerned the undoubted injury to the toe and whether this was caused by disease or by trauma; the probability of accident- in getting out of bed in view of the extent of injury to the toe and the amount of blood spilt; the probability that a man who has a heart attack while in bed *909will not attempt to get out of bed, but will lie down; the probability of injury being connected with a fall rather than with a heart attack; the probability or improbability of receiving such an injury after the onset of the attack and after the patient had already gotten out of bed onto the bare hospital floor; the probability that a fall will precipitate such an attack; the fact that dead men do not bleed, and the probability of one who has already suffered a heart attack being so injured and bleeding copious amounts thereafter during the progress of of the fatal attack, and so on. From all the facts and circumstances the jury have the right to choose the more likely inference; either that the insured injured himself during a fall, thus precipitating the attack, or that he had an attack and thereafter injured himself in any manner the jury would be authorized to find this might have occurred. And it might well be added in support of the Bridges case, supra, that while Judge Powell in Georgia Ry. & Electric Co. v. Harris, 1 Ga. App. 714 (3) (57 S. E. 1076) stated flatly that “an inference resting only upon an inference is not permissible,” he later in Lee v. State, 8 Ga. App. 413, 419 (69 S. E. 310) referred to the statement as “a doctrine, however, of limited applicability, and, indeed, of doubtful soundness.” The trial court erred in directing a verdict in favor of the defendant as contended in special ground 7 of the motion for a new trial.
Error is assigned in ground 4 of the amended motion for a new trial on the ruling of the trial court disallowing an amendment to the petition. Objections to rulings on pleadings are not proper grounds of a motion for a new trial. Kelly v. Strouse & Bros., 116 Ga. 872 (6) (43 S. E. 280). The fourth special ground is without merit.
Whether or not insurance companies take into consideration the condition of the applicant’s heart in issuing accident insurance policies has no relevancy to this case, the issue being whether the death is compensable within the terms of the policy. It was not error to exclude this evidence as set out in special ground 5, and the same is true of the evidence the exclusion of which is assigned as error in special ground 6.
The trial court erred in overruling the motion for a new trial.
*910 Judgment reversed.
Nichols, Bell, Frankum and Jordan, JJ., concur. Felton, C. J., and Carlisle, J., dissent.