Opinion by
Mr. Justice Drew,This action of assumpsit was brought by plaintiff to recover double indemnity upon four policies of life insurance issued by defendant to Dr. Samuel McClary, 3d, the insured, who died on September 20, 1936. The policies, which had a face value of $25,000 in the aggregate, contained identical clauses providing for the payment of double indemnity in the event that “the death of the insured resulted in consequence of bodily injury effected solely through external, violent and accidental means . . . independently and exclusively of all other causes.” Provision was also made that: “This indemnity shall not be payable if the death of the insured results directly or indirectly from disease or from bodily or mental infirmity.” It is upon the effect of these restrictive clauses that defendant bases its rejection of plaintiff’s claim. The face value of the policies, with accrued dividends, has been paid in full.
In the statement of claim plaintiff averred that Dr. McClary died “solely in consequence” of injuries sustained in an accident on August 31, 1936, at Camden, Maine, when his automobile collided with a water hydrant at the side of the highway. The affidavit of defense admitted the happening of the accident, but denied that the insured suffered any injuries resulting in his death. It was alleged that, on the contrary, his death was caused directly or indirectly by a preexisting condition of cancer. Upon this issue the case proceeded to trial.
Evidence was introduced by plaintiff to show that Dr. McClary, a retired physician 59 years of age, was in reasonably good health at the time of the accident. He had suffered a nervous breakdown in 1931, and had received treatments over a period of years for arthritis, sinusitis, and prostate trouble. Nevertheless, it was testified that none of these conditions had severely impaired his health, and that his activities and appearance were normal for a man of his age. He possessed *536a good appetite, took frequent walks, and enjoyed driving Ms own automoMle for recreation. He was alone in the car at the time of the accident, to which there were no witnesses. A chauffeur who was the first to arrive at the scene of the collision testified that he saw Dr. Me Clary rising from the ground a short distance from his automobile in a weak and dazed condition. His speech was thick and difficult to understand, and he put his hands to his head, stating that he was hurt. The witness drove him immediately to- the boarding house at which he was spending his vacation, and then returned to the place where the accident had occurred. Dr. McClary’s automobile was badly damaged, and the condition of it and the hydrant indicated a collision of violent force.
Witnesses who observed the insured at the boarding house after the accident testified that he appeared highly nervous, confused and exhausted. He complained of headaches, and reeled in his walk. His landlady and her sister applied cold compresses to his head during the evening and part of the night. On the following morning, they heard the sound of a fall in Dr. MeClary’s' room, and found him upon the floor in a state of collapse. They put him to bed and summoned his friend and physician, Dr. Codman, from a nearby summer resort. Dr. Codman, who examined the insured on September 2, 1936, testified that he found him suffering from an injux*y to the head, upon which he saw visible bruises, and that he was somnolent, mentally confused, and unable to stand. Dr. McClary’s relatives wex*e informed of his condition, and on September 15th he was brought to Philadelphia, whex*e he was admitted to the Presbyterian Hospital. Five days thereafter he died, following a convulsion. His death was at first ascxdbed to a cerebral hemorrhage or thrombosis, but these causes were excluded by a post mortem examination conducted nine months after, interment. Dr. Codman and a consultant pathologist gave it as their opinion that Dr. *537McClary died as tlie direct result of a concussion of the brain sustained in the accident, and that no other cause existed as a contributing factor. These physicians stated that although Dr. McClary was suffering from chronic myocarditis and arteriosclerosis, these conditions were passive and incidental to the age of the insured.
The defendant offered evidence to rebut plaintiff’s testimony that Dr. McClary had complained of a head injury, and that he had been bruised in the accident. The report and diagnosis of the interne who examined him on his admission to the hospital was introduced for this purpose, as well as the testimony of witnesses who saw him shortly after the collision. The final hospital diagnosis, ascribing Dr. McClary’s death to arteriosclerosis, arthritis, sinusitis, cardiovascular disease and pulmonary edema was also presented. Dr. Moon, a pathologist who assisted in the autopsy, testified that he found the insured emaciated and affected with carcinoma of the liver and a malignant intestinal tumor. He attributed his death to this cause and to a combination of diseases. Dr. Kolmer, plaintiff’s pathologist, had also examined the body of the insured after disinterment, but found no trace of cancer. A medical expert called by the defendant, stated in answer to hypothetical questions that if Dr. McClary had suffered a severe blow on the head in the accident, he would regard it as a contributing factor, but not as a cause, of his death.
Upon this contradictory testimony the case was submitted to the jury by the trial judge, who charged that the burden of proof was upon the plaintiff to show that the insured’s death had resulted solely and exclusively from accidental causes, and that it did not result directly or indirectly from disease or other infirmity. But, at plaintiff’s request, he instructed the jury that the burden was upon the defendant to affirmatively prove its ■defense that the insured died from causes excepted from the operation of the policy. He further charged that an accidental injury should be regarded as the sole, ex-*538elusive and independent cause of death, within the meaning of the restrictive clauses quoted, if it aggravated a preexisting disease and thereby hastened the death of the insured; if it set in motion a chain of events leading to death by operating upon conditions of a passive nature, insufficient in themselves to result fatally; if it was the “moving, sole and proximate cause of the death”, and if it was the “predominant” cause.
The jury returned a verdict for the plaintiff in the sum of $29,995.84, after deliberating for eight hours. Defendant’s motions for a new trial and judgment n. o. v. were overruled by the court en banc, and from the entry of judgment on the verdict this appeal was taken. It is urged that the verdict is opposed to the evidence, that the plaintiff’s testimony failed to sustain its burden of proof, and that the portions of the charge referred to were erroneous.
This Court is of opinion, that plaintiff’s evidence was sufficient to require the submission of the case to the jury upon proper instructions. There was testimony that the insured suffered an injury to the ¡head in the collision, and that his death was caused solely by the concussion. Plaintiff’s medical witnesses excluded disease as a contributing cause. The credibility and weight of this evidence was for the jury to determine. But we are satisfied that a new trial must be awarded because of error in the charge prejudicial to defendant.
The effect of restrictive clauses such as those contained in the double indemnity provisions of these policies has been settled in this State by a series of decisions culminating in certain definite rules of construction. The first of these cases which we shall consider is Kelley v. Pittsburgh Casualty Co., 256 Pa. 1, and the most recent, Lucas v. Metropolitan Life Ins. Co., 339 Pa. 277. In the Kelley case, the policy of insurance contained only a clause limiting recovery to loss or disability “resulting directly, independently and exclusively of all other causes, from bodily injuries effected solely *539through accidental means.” The court below, distinguishing the case from those involving contracts which excluded losses occurring “wholly or partially, directly or indirectly, of any disease, bodily or mental infirmity”, adopted the following quotations as determinative of liability: “ ‘If disease, while existing, be but a condition and the accident the moving, sole and proximate cause of the death, the exception in the policy will not relieve the insurer for death so caused.’” (p. 6.) And, “‘The fact that the physical infirmity of the victim may be a necessary condition to the result does not deprive the injury of its distinction as the sole producing cause. In such case, disease or low vitality do not arise to the dignity of concurring causes, but . . . appear rather as the passive allies of the agencies set in motion by the injury. . . . Where accidental injury aggravated a disease, and thereby hastened death so as to cause it to occur at an earlier period than it would have occurred but for the accident, it is the direct, independent, and exclusive cause of death. . . . The phrase “resulting directly, independently, and exclusively in death,” refers to the efficient, or, as some courts speak of it, the predominant cause of death. . . (p. 7). This language, it is to be observed, is essentially embodied in the charge in the present case. It was held that a verdict for the insured was proper where the evidence showed that he had suffered a fall, which, because of preexisting adhesions, resulted in an intestinal injury. This Court affirmed without opinion. (See also Dale v. Standard Accident Ins. Co., 307 Pa. 398.)
This case was followed by Hesse v. Traveler’s Ins. Co., 299 Pa. 125, which involved a clause similar to that found in the Kelletf case, and contained no additional clause excluding liability for injury or death resulting directly or indirectly from disease or infirmity. A majority of this Court held, three judges dissenting, that death resulting from the proper administration of an anaesthetic due to the insured’s hypersusceptibility, was *540not accidental. The Kelley case was distinguished upon the ground that there the insurer could have “ascertained the existence of the adhesions”, whereas in the Hesse case, the plaintiff’s hypersusceptibility could not have been discovered. It is apparent, however, that the question in the Hesse case was, “Was there an accident?”, while in the Kelley case the question was, “Did the insured’s accident cause his death?”
In the case of Ewing v. Equitable L. A. S. of U. S., 320 Pa. 577, the policy contained a clause excluding liability in the event that loss or death resulted “directly or indirectly, wholly or partly, by bodily or mental infirmity”. There, the insured experienced, following an accident, a recurrence of a nervous disorder from which he had previously suffered. We held that the plaintiff’s existent condition was the effective cause of his illness, and denied recovery. The Kelley case was distinguished upon the ground that it involved a contract which did not include a restrictive clause relating to disease or infirmity. Regarding the charge of the court below we said (p. 580) : “The jury was told that there could be no recovery unless the loss and disability was solely attributable to the accident. . . . All of the evidence was to the effect that the accident was not the sole factor, and a verdict for plaintiff could not have been sustained.”
The next pertinent authority is Arnstein v. Metropolitan L. Ins. Co., 329 Pa. 158. There, the policy in suit ■ contained a clause similar to the ¡one construed in the Ewing case. The insured died as a result of a streptococcic infection originating in a burn which he sustained accidentally. It had been found as a fact in the lower court that although he was afflicted with diabetes, this disease did not contribute to the insured’s death. In affirming the judgment for plaintiff, we said (p. 161) : “Had the death resulted from the combination of the injury and the diabetes, and not from the injury alone, then, even though the injury were the proximate and *541the diabetes merely the remote cause, there would be no liability under the policy.” (Italics added.)
From these two decisions it would appear that the “proximate cause” rule adopted in the Kelley case has no application in construing policies which contained the additional “disease and infirmity” clause which was before us in the Eioing and Arnstein cases. But, in the Arnstein case, this Court adopted an interpretation of the term “bodily infirmity”, as used in the latter clause, consistent with that expressed by the Massachusetts court in Freeman v. Mercantile Mut. Acc. Ass’n, 156 Mass. 351, 30 N. E. 1013; and by Judge Cardozo in Silverstein v. Metropolitan Life Insurance Co., 254 N. Y. 81, 171 N. E. 914. We said (p. 162) : “The latter term in insurance policies has uniformly been construed to mean a condition of a settled and substantial character materially impairing the bodily processes, and not to cover minor physical defects and ailments which are frequent incidents of life, speedily forgotten, and not affecting the general soundness and healthfulness of the system.” In the Silver stein case Judge Cardozo stated: “In a strict or literal sense, any departure from an ideal or perfect norm of health is a disease or an infirmity. Something more, however, must be shown to exclude the effects of accident from the coverage of a policy. The disease or the infirmity must be so considerable or significant that it would be characterized as disease or infirmity in the common speech of men. ... A policy of insurance is not accepted with the thought that its coverage is to be restricted to an Apollo or a Hercules.” The learned jurist supplied the following test: “A distinction, then, is to be drawn between a morbid or abnormal condition of such quality or degree that in its natural and probable development it may be expected to be a source of mischief, in which event it may fairly be described as a disease or an infirmity, and a condition abnormal or unsound when tested by a standard of perfection, yet so remote in its potential mischief that *542common speech would call it not disease or infirmity, but at most a predisposing tendency.” (Italics added.)
From the Arnstein case, therefore, it was clear that a plaintiff suing upon a policy of insurance containing both of the restrictive clauses which are here considered, had the burden of showing that the loss or death of the insured was caused solely by accidental means, and that it was not produced directly or indirectly by disease or infirmity, but that recovery was not dependent upon establishing that the insured was in a state of perfect health and free from normal ailments and predispositions compatible with his age. This rule was reiterated in Gyulai v. Prudential Ins. Co. of Am., 135 Pa. Superior Ct. 73, and applied to a case in which the policy contained no clause referring to disease or infirmity. There it was held that the hypersensitivity of the insured to anti-tetanus serum could not be regarded as a “disease or bodily infirmity”, in the common acceptance of the terms.
This was the state of the law when the case of Kelly v. Prudential Ins. Co., 334 Pa. 143, was decided by this Court. The facts of that case are deserving of special attention. The insured, Mrs. Catherine Daley, took out policies of life insurance providing for double indemnity in the event of accidental death, at the age of 53. The policies contained both, of the restrictive clauses. Seven years later she fell upon an unlighted flight of cellar steps and sustained severe injuries to the head which resulted in her death. At the time she was suffering from an arteriosclerotic condition not unusual in persons of her age. At the trial, two issues were presented: “first, whether the fall was the result of her tripping on the steps, and hence of an, accident, or was occasioned by an attack of vertigo or dizziness incidental to a chronic condition of arteriosclerosis; and, second, whether, if the fall was accidental, the death which ensued was the result of the fall alone or of the combined effect of the fall and of the preexisting ar*543teriosclerotic condition.” Upon both questions, the jury found for the plaintiff. This Court affirmed the judgment, holding that there was sufficient evidence to support the verdict of the jury.
There is clearly, therefore, no conflict between the direct holding in the Kelly case, upon its facts, and our decisions in the Ewing and Arnstein cases. The only apparent divergence lies in the approval expressed by this Court of certain portions of the charge of the court below. The trial judge, ignoring or failing to recognize the distinction between the policy before it, containing a “disease and infirmity” clause, and those having no such restriction, charged the jury in the language of the Kelley case that if the accident was the “proximate cause” of the insured’s death, it was to be regarded as the “direct, independent, and exclusive” cause within the meaning of the policy. It was held, however, that this statement of the law was not opposed to that expressed in the Ewing and Arnstein cases, and its effect was stated to be as follows (p. 151) : “It is clear that if the physical condition of the insured is merely weakened or his resistance to disease lowered by preexisting ailment or disease incident to advancing age, which creates a bodily condition of a passive nature not alone sufficient to cause death, this fact will not prevent recovery . . . even though death resulting indirectly from bodily or mental infirmity or disease is excluded as a risk, provided the accident sets in progress the chain of events leading directly to death.” (Italics added.) So understood, of course, the language of the charge in the Kelly case is not greatly at variance with the rule asserted in the Arnstein case, despite the unfortunate reference of the trial judge to “proximate” and “predominant” causation. More difficult to reconcile, however, is the refusal of defendant’s point for charge in the Kelly case, which read: “There can be no recovery in this case for ‘Double Indemnity’ unless the jury find, (a) that the death of the insured was due to acci*544dental means; (b) that the death was due solely and entirely to accidental means, and not contributed to by disease; (c) and that it was not due, either directly or indirectly, by disease or bodily infirmity.” If the words “disease or bodily infirmity” are given the meaning adopted in the Arnstein case, this request was a correct statement of the effect of the restrictive clauses of the policy. It was objectionable only in that it did not define these terms within the limitations there expressed. Our affirmation of the refusal of this point must*be considered therefore in the light of the facts of the case, the language of the general charge, and our reassertion of the rule embodied in the Ewing and Arnstein decisions, which were clearly not overruled or criticised.
Following the Kelly case, it was held in Roeper v. Monarch Life Ins. Co., 138 Pa. Superior Ct. 283, in a suit upon a similar contract of insurance, that the plaintiff was required to show that the disability and death of the insured resulted solely from accidental injury. There, the insured suffered an injury to his foot by accidental means. Gangrene set in and an amputation was thereby necessitated. After the operation the insured was discharged from the hospital as a convalescent, but died within a few days as the result of a coronary occlusion. The plaintiff’s witnesses testified that the insured was suffering from diabetes, and that gangrene was of a diabetic character. None of them excluded diabetes or arteriosclerosis as a cause of death. In holding that a judgment should have been entered for defendant at the close of plaintiff’s testimony, President Judge Keller said (p. 287) : “While the injury to the foot may have set in motion forces which ultimately caused the disability and death, the diabetes unquestionably contributed to the gangrene, which necessitated the amputation of the foot, and the arteriosclerosis was a contributing factor to the coronary occlusion, which, following the amputation, caused the insured’s death.” (Italics added.)
*545All of the foregoing eases were recently considered by this Court in Lucas v. Metropolitan Life Ins. Co., supra, where, as in the Roeper case, the plaintiff was denied recovery because his evidence did not exclude the possibility that a preexisting, substantial infirmity contributed to the -death of the insured, “but provided an affirmative basis for such a conclusion.” We said (p. 280): “Where the liability of the insurance carrier is so restricted, it is not sufficient for the insured to establish a direct causal relation between the accident and the loss or disability. He must show that the resulting condition was caused solely by external and accidental means, and if the proof points to a preexisting infirmity or abnormality which may have been a contributing factor, the burden is upon him to produce further evidence to exclude that possibility.” (Italics added.) We further pointed out that preexisting condition which contributed to the death of the insured was not a normal or usual incident of advancing age, as in the Kelly case, but “a severe and abnormal infirmity”. The Ewing, Arnstein and Silverstein cases were cited with approval.
Upon an analysis of these authorities we are of opinion that the charge of the court in the present case was misleading to the jury. Although the trial judge correctly stated, preliminary to his discussion of the case, that the plaintiff had assumed the burden of proving that the insured “died in consequence of an injury effected solely through external violent and accidental means . . . and that death resulted independently and exclusively of all other causes”, as well as that “this death did not result directly or indirectly from disease or from bodily or mental infirmity”, the effect of this instruction was destroyed by subsequent portions of the charge. The jury was led to believe that if the injury sustained by the insured in the collision was the “primary” or “proximate”, or “predominant” cause of death, or if it “hastened his death”, then regardless of *546any preexisting infirmity of the insured, the plaintiff was entitled to recover. The statement, “If he died as a result of his other infirmities, or if any of them was the primary or immediate cause of his death, then there must be a verdict for the defendant . . [Italics added] would indicate to the jurors that plaintiff could recover although an existing infirmity of the insured merely operated, directly or indirectly, as a contributing cause. In the absence of proper instructions as to the meaning to be ascribed to the words “disease” and “infirmity” in the restrictive clauses, the jury was unquestionably bound to be confused by the apparent contradiction in the judge’s first and subsequent statements of the law.
The charge was furthermore erroneous in its instruction that the burden of proof rested upon the defendant to show that the death of the insured was due to the preexistence of disease. The plaintiff’s own evidence established that the insured was suffering from arthritis, arteriosclerosis and a cardiac condition, and under the rule expressly stated in the Lucas case, the burden was upon it to exclude the possibility that these conditions contributed to the death, or to show that they were not “infirmities” or “diseases” within the meaning of the restrictive clauses. For these reasons a new trial must be awarded.
Counsel for appellant have called to our attention a large number of authorities in other jurisdictions upon the subject here considered, in some of which views have been expressed contrary to the doctrine of the Silver-stein case and to our own decisions. As these authorities are merely illustrative of the great variety of divergent opinion upon the subject,1 and as our own rule is firmly established by the cases above cited, it would serve no purpose to review or discuss them here.
The judgment of the court below is reversed with a venire facias de novo.
See Note, 88 U. of Pa. L. Rev. 853.