EDWARD S. SMITH, Circuit Judge:

In this accidental death benefits insurance case appellant (Barbara T. Collins, executrix and beneficiary of the deceased's, James T. Collins, estate) appeals from a judgment of the United States District Court for the Northern District of Alabama, Eastern Division, denying her accidental death benefits under the policy which her husband the deceased held with appellee (Metropolitan Life Insurance Company). We affirm.

Issues

The principal issue is whether the court below (Judge Propst) erred in refusing to charge the jury with the "chain reaction" theory of recovery under the "additional clause" of the subject insurance policy, in accordance with Alabama law. The secondary issue is whether, if the court so erred, the error is harmless.

Background

James T. Collins was employed as Food Administrator at the Federal Correctional Institution in Talladega, Alabama, on October 15, 1980, the date of his death. Sometime after 4 p.m. that day one of the institution's food service employees, intending to clean the kitchen floors, combined Clorox and Lime-Away with warm water in a mop bucket, only to find that the mixture began to bubble and emit fumes. Alarmed, the employee ordered an inmate to take the mixture through the back door and outside the building. Unfortunately for Mr. Collins, who was leaving work at approximately this time, his path home took him through the kitchen serving line and out the back door. Soon after, he was taken to the dispensary and from there to the local hospital, where he was pronounced dead on arrival. Another inmate who was in the room where the offending chemicals were mixed was also taken to the dispensary, having complained of difficulty in breathing.

Mr. Collins was an asthmatic. He had begun having asthma attacks requiring hospitalization in 1968 when he lived in Hopewell, Virginia. Subsequently in Carbondale, Illinois, he had been hospitalized 4 times and treated in the emergency room approximately 10 times for his condition. Upon his transfer to Talladega in September 1979 he came under the care of Dr. Jimmy Davis and was taking at least four different medications for asthma at that time. In August 1980 Mr. Collins' condition had worsened to the point where he was again hospitalized. Tests then performed led Dr. Davis to conclude that Mr. Collins had "chronic obstructive pulmonary disease," which is irreversible. Dr. Davis also concluded that, even though Mr. Collins' health improved following the hospitalization, his disease was still such that he was at risk to die any time, even without inhalation of a known irritant. In fact, in early October 1980 Mr. Collins had had an asthma attack for which he was treated at work, and his co-workers recognized his constant breathing problems and frequent use of an inhaler for relief. Mr. Collins' death certificate, which Dr. Davis completed, stated that the cause of death was "Cardio Pulmonary Arrest due to, or as consequence of, Asthma."

In November 1980, Mr. Collins' body was exhumed and Dr. Wendell Sowell performed an autopsy. He concluded that the cause of Mr. Collins' death was chlorine poisoning. Dr. Sowell, who had been informed prior to the autopsy of the deceased's inhalation of chlorine fumes, was not a medical doctor nor a pathologist and did not perform a microscopic evaluation or a chemical analysis for evidence of bronchial asthma. He recognized that at the time he performed the autopsy (6 weeks after embalming and burial) there was no way to determine whether chlorine was present in the body. Dr. Davis, upon reviewing the autopsy findings, stated that nothing in those findings caused him to change his conclusion that Mr. Collins had died of cardiopulmonary arrest due to, or as a consequence of, asthma. Dr. Ben Branscomb, a specialist in lung diseases who reviewed Mr. Collins' extensive medical records, testified that he disagreed with Dr. Sowell's autopsy findings and that the conclusion of Dr. Davis as stated on the death certificate was consistent with the deceased's medical history.

Appellee paid appellant $32,000 in natural death benefits but refused to pay an additional $32,000 in accidental death benefits. Under the terms of the deceased's insurance policy those benefits were recoverable if he died as a result of having "sustained bodily injuries solely through violent, external and accidental means, and (died) * * * as a direct result of such bodily injuries independently of all other causes" (the so-called "general clause"), but the benefits were not recoverable if the death was "caused wholly or partly, directly or indirectly, by disease or bodily or mental infirmity" (the so-called "additional clause"). Appellant sued appellee for recovery of the accidental benefits and the jury returned a verdict for the insurance company. The trial court, in charging the jury, refused to grant appellant's requested instructions concerning the "chain reaction" theory of recovery, on the grounds that Mr. Collins did not suffer from a "dormant" disease.

Discussion

A long-standing allegorical tug-of-war exists between the insurance companies and the Alabama courts in the accidental death benefits area, with the insured or insured's beneficiary being a highly partisan spectator. At the one end, the insurance companies have pulled for an accidental death, as defined in their "general clause," which is the equivalent of a truck dropping from the skies, striking squarely and killing instantly a perfectly fit human specimen clutching a just-issued physician's clean bill of health. When the Alabama courts responded to this position with a strong tug the opposite way by reading into the "general clause" a proximate cause test, allowing benefits where, but for an infirmity, the insured would not have died from the accident, the companies pulled back by citing in their policies an "additional clause" designed to negate the courts' undermining (from the companies' point of view) of the general clause. The additional clause disallows recovery where death results from the combined effects of an accident and a pre-existing disease which was accelerated and aggravated by the accident. Metropolitan Life Ins. Co. v. Nichols, 393 So.2d 966, 967 (Ala. 1981). Responding at this point to a negative reaction from the bystanders, however, the Alabama courts have refused to allow the companies to succeed in regaining their original ground, by tugging back at the additional clause with the "chain reaction" theory, which the parties have placed at issue here.

The chain reaction theory is stated as follows:

Alabama courts, and this Court, take a more reasonable view [regarding the additional clause]. If an insured has an active disease of such a character as to endanger the insured's life, apart from the accident, such a disease is a contributing cause that will bar recovery. If however an injury starts a chain reaction resulting in death, recovery may be allowed even if one of the links in the chain is old age frailty and some of the links are dormant diseases or physical conditions without which the chain would be broken. * * * [Emphasis supplied.]

New York Life Ins. Co. v. McGehee, 260 F.2d 768, 773 (5th Cir. 1958). See also Liberty Nat'l Life Ins. Co. v. Reid, 276 Ala. 25, 33, 158 So.2d 667, 674 (1963); Nichols, 393 So.2d at 968.

A.

In determining whether the trial court erred in refusing to charge the jury with instructions concerning this theory, it is our duty as reviewing court to satisfy ourselves that the instructions show no tendency to confuse or mislead the jury regarding applicable principles of law. Green v. Edmands Co., 639 F.2d 286, 289 (5th Cir. 1981). It is improper for a court to instruct on a proposition of law about which there is no evidence. Jackson v. Southern Railway, 317 F.2d 532, 537 (5th Cir.), cert. denied, 375 U.S. 837, 84 S.Ct. 77, 11 L.Ed.2d 65 (1963). We therefore must determine whether there is no evidence in the record to support a chain reaction theory charge, or whether such a charge would only confuse or mislead the jury regarding Alabama law.

In analyzing the chain reaction theory, both the trial court and the parties in oral argument before this court focused almost exclusively on the term "dormant." This focus is certainly appropriate, but it will be useful as well to consider the chain reaction theory in light of the "tug-of-war" context we have outlined above.

Regarding the term "dormant," Judge Propst, in discussing the requested instructions with counsel for appellant and appellee (then plaintiff and defendant), stated his conclusion that the charge was not appropriate because "there is no indication that the underlying disease [asthma] is dormant" and it "is just beyond comprehension to me that it [the asthma] could be considered dormant under any testimony." Appellant counters that the evidence supports the proposition that Mr. Collins' asthma was indeed dormant, since he was able to work full-time, was planning to take his wife out to dinner the evening of his death, and had not been hospitalized with an asthma attack for 2 months.

Our examination of the record and Alabama cases concerning this point leads us to a conclusion agreeing with the district court. While Mr. Collins may not have been hospitalized for 2 months prior to his death on October 15, the record shows that in early October he had suffered an asthma attack at work for which he was treated by the staff physician. Mr. Collins was under constant medication for his disease and was seen by his co-workers frequently using inhalers and oral sprays. Perhaps most importantly, his own personal physician had concluded that his disease was such that he was at risk to die any time. Given these facts, Judge Propst properly concluded that no evidence existed in the record to support a finding of dormancy. Put another way, no reasonable person could conclude that the deceased did not have "an active disease of such a character as to endanger the insured's life, apart from the accident." McGehee, 260 F.2d at 773.

Examination of the limited Alabama case law applying the chain reaction theory in the context of dormant diseases supports this conclusion. In Nichols, supra, the court found that instructing the jury regarding the theory was proper, where testimony established that the underlying disease, a Berry aneurysm which was congenital in nature and had no history of prior leakage, would be termed dormant. Id., 393 So.2d at 968. In McGehee, supra, the court found that substantial evidence supported a jury verdict awarding accidental death benefits to a 77-year-old insured who died from complications occurring after a fall, even though the insured had numerous diseases. Each of the insured's diseases, however, was either inactive or normal for a man of his age: his diabetes had not necessitated treatment for 4 years; his arthritis was only in one hand; his arteriosclerosis was moderate for his age; his heart trouble had not required digitalis for 5 or 6 months. Id., 260 F.2d at 773. Perhaps most importantly, Mr. McGehee's personal physician had testified that McGehee could expect to live another 10 years, and the physician performing the autopsy testified that McGehee had the body of a "middle-aged man." Id. This contrasts sharply with 49-year-old Mr. Collins, who was at risk to die any time. See also Reid, 276 Ala. at 33, 158 So.2d at 674, in which the court affirmed judgment for the insured where medical testimony indicated that the insured's underlying heart disease was not a "cooperating and efficient cause of death." But see Wilson v. Liberty Nat'l Life Ins. Co., 331 So.2d 617 (Ala. 1976), in which the court reversed a summary judgment for the insurer on the grounds that there existed a scintilla of evidence that the insured's heart condition might be termed a dormant disease without which the chain reaction leading to death would have been broken.

Though the evidence in appellant's case establishes that her husband's disease was not dormant but highly active, it is nevertheless useful to examine whether instructions on the chain reaction theory should obtain, in light of the overall "tug-of-war" history concerning the general and additional clauses. Regarding the general clause, which on its face limits recovery to death resulting solely from an accident, Alabama law would allow appellant to recover if the accident were merely the proximate cause of the insured's death. A jury might well have awarded appellant her double indemnity if Mr. Collins' policy with appellee had contained only this clause. However, appellee had "tugged back" by inserting the additional clause in the policy which Mr. Collins had purchased, and under Alabama law the jury could not award the accidental benefits if it found that the chlorine gas accident and the asthma combined together to cause the death, and the asthma was a cooperating and efficient cause of death. Nichols and Reid, supra. The jury found affirmatively regarding the asthma and denied the benefits. Should appellant have been allowed the aid of another judicial "tug" her way in the form of chain reaction theory instructions to the jury? It is possible that, on the additional clause instructions alone, the jury could have found the asthma not to be a cooperating and efficient cause of death and awarded Mrs. Collins the double indemnity, as the jury did in Reid, supra, after hearing conflicting medical testimony about Mr. Reid's heart condition, and in McGehee, supra, after hearing testimony about Mr. McGehee's infirmities. Indeed, when one ponders the Alabama cases herein and the "tug-of-war" relationship among the various actors, one sees that the "chain reaction" theory is, for all practical purposes, the negative of the additional clause "combined effects" or "cooperating and efficient cause" theory. In other words, a jury finding that the insured's disease was not an efficient and cooperating cause of death is in effect a finding that the disease is dormant. While we certainly do not intend by this analysis to write the chain reaction theory out of Alabama law, we do aim to analyze it in context. That context is as a last judicial tug on behalf of the insured, where the evidence so merits. Here we find that the evidence does not so merit, and that to have instructed the jury concerning the theory would have confused and misled the jury members. We therefore affirm the trial court's judgment.

B.

Since we have affirmed the court below on the merits of not instructing the jury regarding the chain reaction theory, we need not discuss the question whether failure so to instruct was harmless error.

Conclusion

Having examined the record and relevant Alabama case law, we conclude that the highly active nature of the deceased's disease was such that the trial court did not err as a matter of law in refusing to instruct the jury regarding the chain reaction theory of recovery under Alabama law. The district court's judgment denying appellant accidental death benefits is therefore affirmed.

AFFIRMED.