delivered the opinion of the Court.
Mario Santoni, age 56, died on July 3, 1972, of hepatitis caused by his taking of the drug isoniazid (INH). The drug was dispensed to him in a tuberculosis prevention program conducted by the Baltimore City Health Department. As a result, on April 29,1975, Antoinette Santoni, his widow and the personal representative of his estate, brought a medical malpractice suit under the wrongful death statute against Dr. Allan S. Moodie, Director of the Bureau of Communicable Diseases of the Baltimore City Health Department and head of the Baltimore City tuberculosis program; Dr. Meyer W. Jacobson, Clinical Director of the Baltimore City tuberculosis program; and Dr. Herman Schaerf, a clinical doctor *500in the Baltimore City tuberculosis program.1 A trial of the case was held in the Superior Court of Baltimore City in February, 1980. The trial consumed more than two weeks and involved the testimony of thirteen witnesses. The case was eventually submitted to the jury on the issues of the defendants’ negligence and the decedent’s contributory negligence. On February 29, 1980, a jury rendered a verdict finding no negligence on the part of Dr. Schaerf which caused or contributed to the decedent’s death but negligence on the parts of Drs. Moodie and Jacobson. The jury also found, however, that the decedent’s own negligence caused or contributed to his death, thereby barring recovery by the plaintiff. After Mrs. Santoni’s motion for judgment n.o.v. and for a partial new trial or a new trial was denied, she filed this appeal. The defendants have filed a cross-appeal.
On this appeal, the appellant raises the following contentions:
1) That the decedent was not contributorily negligent as a matter of law and that the trial court, therefore, erred in permitting the question of his contributory negligence to go to the jury;
2) That the trial court erred in its instructions on contributory negligence;
3) That the trial court committed reversible error in refusing to admit into evidence a certain conversation between the appellant and her husband;
4) That the trial court committed reversible error in admitting into evidence a "Danger Signs” poster; and
5) That the jury’s verdict was inconsistent and irreconcilable..
The defendant-appellees, on the other hand, in their cross-appeal contend that their motions for a directed verdict *501on the issue of immunity should have been granted. They also contend that the court erred in admitting evidence of twelve other deaths of persons who received INH therapy from the Baltimore City Health Department.
The drug isoniazid (INH) has been used since the early 1950’s for the prevention and treatment of tuberculosis. Its metabolism is by the liver. Until the mid-1960’s, it was thought to be one of the safest drugs on the market. In 1970, however, after two men who were participants in a tuberculosis prevention program in Washington, D.C., died from hepatitis, it was recognized that the ingestion of INH might be associated with liver disease, particularly hepatitis. As a result, in 1971, the United States Public Health Service recommended that certain screening and monitoring procedures be used on all recipients of the drug in any INH program. The Service also recommended that a study be made of persons taking INH to determine if there were any connection between the INH and liver disease. Following up on its recommendation of a surveillance program, the Public Health Service sought to enlist the help of local health departments to report on the progress of patients in local health department INPI programs. The Service would keep the statistical information, while the local public health agencies would dispense the drug and monitor the people. Baltimore City at that time and for many years before had one of the highest tuberculosis rates in the country, both in number of cases and deaths. In order to expand its already existing prophylaxis program the Baltimore City Health Department agreed to participate in the special surveillance program.
The protocol for the surveillance program promulgated by the United States Public Health Service called for dispensing the drug, in pill form, only on a monthly basis. A patient would thus be required to return to the clinic once a month to receive his 30-day supply of pills. At the time of these visits, any adverse reactions would be noted, particularly any possible signs and symptoms of hepatitis. The protocol provided that the participant would be "ask[ed] specifically about jaundice, dark urine, sustained loss of *502appetite, marked fatigue and prolonged nausea or vomiting.” In any cases of suspected hepatitis, the isoniazid would be discontinued immediately. A monthly report card would be sent to the Public Health Service on each patient.
In 1972, Dr. Allan S. Moodie was the Administrative Health Officer in the Baltimore City Health Department in charge of the administrative aspects of the isoniazid program such as arranging funding and staff for the program. He arranged for the Baltimore City Health Department to participate in the United States Public Health Service surveillance program. Dr. Meyer W. Jacobson was the Clinical Director of the Baltimore City tuberculosis program. Eventually approximately 3,000 persons entered the surveillance program.
The program in Baltimore City provided that a clinic doctor would make the determination whether a person would be placed on INH therapy, based on the tuberculin skin test and chest X-ray. Thereafter, a nurse would instruct the patient on the use of the drug and any signs and symptoms of a reaction. At the follow-up monthly visits, at first only public health nurses were permitted to interview the participants in the program. As the program expanded, however, health aides and clerical help were authorized to conduct the interviews. If a patient reported any side effects, the patient was to be referred to a clinic doctor for further examination. Any reaction to the drug was to be reported on a chemoprophylaxis register sheet and on the computer cards sent monthly to the Public Health Service.
Mario Santoni became a participant in the Baltimore City isoniazid program in 1972. Mr. Santoni, a native of Italy, came to the United States in 1955. Two years later, he married the appellant, Antoinette Santoni. In Italy, Mr. Santoni had completed only the fourth or fifth grade and, in this country, worked primarily as a painter and carpenter. In 1972, he spoke only broken English and had great difficulty reading it. Prior to the illness which caused his death, his health had been excellent and, according to his wife, he had never been to a doctor for any illness.
*503In January, 1972, Mr. Santoni became a naturalized citizen of the United States and also became employed by Baltimore City as a maintenance man at the Southern District Police Station. As a prerequisite to his employment, he was required to undergo a medical examination with a tuberculin skin test and chest X-ray. The examination was done at the Eastern Health District Chest Clinic of the Baltimore City Health Department. It showed a positive tuberculin skin test but a clear chest X-ray. This combination of a positive skin test and clear chest X-ray indicated that Mr. Santoni did not have active tuberculosis but that he had live tubercle bacilli in his body. It indicated that he had a greater than ordinary risk of developing tuberculosis.
Dr. Herman Schaerf reviewed the test and X-ray and, on January 20, 1972, placed Mr. Santoni in the Baltimore City isoniazid program. The program called for Mr. Santoni to receive the drug isoniazid prophylactically for one year. Although isoniazid is a prescription drug, no physician in the program ever saw Mr. Santoni, and no informed consent form to utilize the drug was ever signed by Mr. Santoni or any of the approximately 3,000 other people who were in this special surveillance program.
Mr. Santoni faithfully went to the clinic each month to receive his monthly supply of the medication. He believed that in order to work for the City he had to take the isoniazid and regularly took his pills. In March, 1972, Mr. Santoni began to experience fatigue and decreased appetite. By April, he noted fullness, abdominal discomfort, and increased flatulence. By May, fatty food intolerance became evident, his stools became lighter, and his urine became darker. By early June, Mr. Santoni began to look worse and to feel more tired. His final visit to the clinic was on June 12, 1972.
During the month of May, Mr. Santoni began to look more pale and began to experience more gastric discomfort. His wife was away for a month at the shore with other members of her family. He complained about his condition to a neighbor. When Mrs. Santoni returned home in mid-June, she became concerned about her husband’s condition and *504consulted her doctor. The doctor prescribed cbantigel and librium for her husband’s stomach problems and nerves. When Mr. Santoni’s condition became worse, Mrs. Santoni called Dr. Joseph Notarangelo at Mercy Hospital. Mr. Santoni was admitted to the hospital on June 22, 1972.
At the time of Mr. Santoni’s admission to Mercy Hospital, he complained of excessive gas and indigestion and gave a fairly detailed history of his illness. He did not, however, know that he was jaundiced. At the hospital, Mr. Santoni was diagnosed as suffering from hepatitis. There is no treatment for the disease, only care to prevent complications. Despite the absence of any complications, Mr. Santoni’s condition deteriorated rapidly and he died on July 3,1972. Expert testimony placed the cause of death as a toxic reaction to isoniazid, causing hepatitis. Twelve other participants in the tuberculosis prevention program also died of liver disease in 1972.
There was expert testimony that had the disease been diagnosed in early June, Mr. Santoni’s chances for recovery would have increased significantly. There was further testimony that Mr. Santoni was probably jaundiced at the time of his last clinic visit on June 12. The chemoprophylaxis register sheet, which contained a history of Mr. Santoni’s visits to the clinic all along, however, made no mention of any adverse signs or symptoms.
It was not until the fall of 1972 that Baltimore City officials first became aware of the large number of deaths of participants- in the City isoniazid program. It was then brought to light that there were thirteen deaths in 1972 of participants in the program (both the general program and special surveillance program combined) and that one of these (Mario Santoni’s) was directly related and seven were probably related to the INH therapy. Five were listed as possibly related to INH therapy. The City INH prophylaxis program was thereafter severely restricted. Isoniazid was prescribed as preventive therapy only for certain high-risk individuals.
In addressing the specific contentions raised in this appeal and cross-appeal, we note at the outset that the question of *505negligence is not in issue here. That question was decided by the jury adversely to the appellees and was not raised in their cross-appeal. We are concerned only with the question of contributory negligence — whether or not there was legally sufficient evidence of contributory negligence on the part of Mario Santoni to support the jury’s finding. If there were not, a directed verdict should have been granted on this issue and the jury should never have considered the question.
Contributory negligence is a recognized defense in medical malpractice cases. It bars recovery where the patient’s negligence was an active and efficient contributing cause of the injury. "The rule of contributory negligence requires that the patient’s negligence must be concurrent with that of the physician. If it occurs after the physician’s negligence and merely adds to the effects, as opposed to being the cause of the patient’s problem, it will not relieve the physician from liability; it will merely serve to 'mitigate’ or lessen the amount of damages awarded to the patient.” Holder, Medical Malpractice Law, p. 302 (2nd ed. 1978).
The burden of establishing contributory negligence was on the defendants. In Batten v. Michel, 15 Md. App. 646, 652, 292 A.2d 707, we said:
"Contributory negligence is an affirmative defense and the burden of proving the plaintiffs contributory negligence rests upon the defendant. . .. Contributory negligence, if present, defeats recovery because it is a proximate cause of the accident....” (Citations omitted.)
A person is contributorily negligent when he fails to exercise ordinary and reasonable care for his own safety by doing something that a person of ordinary prudence would not do or failing to do something that a person of ordinary prudence would do, under the circumstances. In Menish v. Polinger Company, 277 Md. 553, 559, 356 A.2d 233, the Court of Appeals discussed this standard of care:
"In measuring contributory negligence, the standard of care to be used as the criterion is that of an *506ordinarily prudent person under the same or .similar circumstances, not that of a very cautious person. Sanders v. Williams, 209 Md. 149, 153, 120 A.2d 397, 399 (1956); and what an ordinarily prudent and careful person would do under a given set of circumstances is usually controlled by the instinctive urge of one to protect himself from harm. Greer Lines Company v. Roberts, 216 Md. 69, 79, 139 A.2d 235, 239 (1958); Martin v. Sweeney, 207 Md. 543, 548, 114 A.2d 825, 827 (1955).”
An important element of contributory negligence is the foreseeability of harm. To be held contributorily negligent, a person must actually have been aware of or should have appreciated the risks involved and then failed to exercise reasonable and ordinary care for his own safety. The courts have on many occasions discussed this element of foreseeability of risks and failure to exercise ordinary care. In Menish v. Polinger Company, supra, the Court of Appeals, stated, at 277 Md. 560-561:
"Before the doctrine of contributory negligence can be successfully invoked, it must be demonstrated that the injured party acted, or failed to act, with knowledge and appreciation, either actual or imputed, of the danger of injury which his conduct involves. . . . Stated another way, when one who knows and appreciates, or in the exercise of ordinary care should know and appreciate, the existence of danger from which injury might reasonably be anticipated, he must exercise ordinary care to avoid such injury; when by his voluntary acts or omissions he exposes himself to danger of which he has actual or imputed knowledge, he may be guilty of contributory negligence.” (Citations omitted.)
In Sanders v. Williams, supra, the Court stated, at 209 Md. 152:
"As is true of primary negligence, one measure of contributory negligence is the need, in a given sit*507uation, to anticipate danger. Presence or absence of reasonable foresight is an essential part of the concept. One is charged with notice of what a reasonably and ordinarily prudent person would have foreseen and so must foresee what common experience tells may, in all likelihood, occur, and to anticipate and guard against what usually happens. On the other hand, one is not bound to anticipate every possible injury that may occur or every possible eventuality.... Absent actual or constructive knowledge to the contrary, one may act on the assumption that he will not be exposed to danger that will come only by the breach of duty which another owes him. He is not bound to anticipate negligent acts or omissions on the part of others unless, under the circumstances, an ordinarily prudent person would know, or should know, that it was not safe to make the assumption of due care on the part of the other person.” (Citations omitted.)
With respect to this element of foreseeability, in discussing contributory negligence in medical malpractice cases, courts have noted the disparity between the knowledge and skill of a doctor and that of a patient. The patient is not in a position to diagnose his own ailment. Without being told, he does not know the risks of medication. He is not in a position to judge whether the prescribed course of treatment is in his best interest. As a consequence, it is not contributory negligence for a patient to follow a doctor’s instructions or rely on his advice, Largess v. Tatum, 130 Vt. 271, 291 A.2d 398 (1972), Los Alamos Medical Center, Inc. v. Coe, 58 N.M. 686, 275 P.2d 175 (1954); to fail to consult another doctor when the patient has no reason to believe that the doctor’s negligence has caused his injury, Johnson v. United States, 271 F. Supp. 205 (W.D. Ark. 1967), Rahn v. United States, 222 F. Supp. 775 (S.D. Ga. 1963), Fairchild v. Brian, 354 So. 2d 675 (La. App. 1977); or to fail to diagnose his own illness, O’Neil v. State, 66 Misc. 2d 936, 323 N.Y.S.2d 56 (1971). See also Martineau v. Nelson, 311 Minn. 92, 247 N.W.2d 409, 417 (1976); Annot., Contributory Negligence or *508Assumption of Risk as Defense in Action against Physician or Surgeon for Malpractice, 50 ALR2d 438.
The patient has a right to rely on the doctor’s knowledge and skill. The court in Halverson v. Zimmerman, 60 N.D. 113, 232 N.W. 754, 759 (1930), stated well the nature of this reliance:
"It is not a part of the duties of a patient to distrust his physician, or to set his judgment against that of the expert whom he has employed to treat him or to appeal to other physicians to ascertain if the physician is performing his duty properly. The very relation assumes trust and confidence on the part of the patient in the capacity and skill of the physician; and it would indeed require an unusual state of facts to render a person who is possessed of no medical skill guilty of contributory negligence because he accepts the word of his physician and trusts in the efficacy of the treatment prescribed by him. A patient has the right to rely on the professional skill of his physician, without calling others in to determine whether he really possesses such skill or not. The patient is not bound to call in other physicians, unless he becomes fully aware that the physician has not been, and is not, giving proper treatment.”
57 Am. Jur. 2d, Negligence, § 324, discusses the comparative knowledge of the parties as bearing on the question of liability.
"Since knowledge of danger, either actual or imputed, is an element of negligence, whether primary negligence on the part of the defendant, or contributory negligence on the part of the plaintiff, the question of liability is sometimes resolved in a negligence action by a comparison of the knowledge of the defendant as against the knowledge of the person injured. The proposition may be stated as follows: Liability is established when it is shown that the peril, being of the defendant’s creation, was *509known to the defendant but not to the person injured; but no liability is predicable of the injury when it appears that the injured person’s knowledge of the danger surpassed or equaled that of the defendent.”
Applying that body of law to the evidence, or lack thereof, in the case at hand, we note that there was no direct evidence at trial that Mario Santoni knew of the risks of taking INH. No informed consent form was ever signed by him nor was there any notation on his chemoprophylaxis register sheet that he was advised of the risk of hepatitis. Whatever evidence was introduced on this issue was totally circumstantial.
The appellees sought to establish through a series of inferences that Mr. Santoni was aware of the risks associated with isoniazid. The evidence showed that the protocol set forth by the U. S. Public Health Service for questioning a patient about the appearance of any adverse signs or symptoms was distributed to all the health clinics in Baltimore City. This protocol was discussed at the monthly meetings of the charge nurses from all the clinics. The administrative nursing supervisor assigned to the Division of Tuberculosis Control testified that she sat in on some of the interviews and saw the procedure being followed. From this evidence, the appellees argue that the inference could be made that Mr. Santoni was aware of the risks of isoniazid. From that inference, they then contend another inference could be drawn — that the absence of any notation on the register sheet as to any adverse signs or symptoms was due to Mr. Santoni’s failure to advise the clinic personnel of these reactions. As a consequence, they urge that the jury could have made the ultimate inference that Mr. Santoni was negligent in continuing to take the INH after the appearance of his symptoms and through his failure to seek immediate medical attention was responsible for his death.
To be sure, from the defendants’ evidence itself, as well as from additional evidence on the issue offered by the plaintiff, a strong countervailing argument could also be mounted. A circumstantial predicate was established that arguably *510showed that Mr. Santoni did not indeed know of the risks of isoniazid ingestion or associate his symptoms with the medication because he was not properly questioned about adverse reactions. Mr. Santoni was an immigrant with little formal education. He and his wife spoke Italian at home. He spoke only broken English and had great difficulty reading English. He did not go to doctors for any medical treatment. Under the impression that he had to take the INH to retain his job with the City, he went to the clinic regularly to get the medication. He had a right to rely on the clinic personnel in his treatment. They were in a far superior position to recognize any adverse signs or symptoms. Mr. Santoni first began to notice ill-defined symptoms of fatigue and decreased appetite in March. The evidence indicates that he continued to associate his symptoms with stomach problems, as is evident on his hospital admission record. At the time of his admission, he complained of excessive gas and indigestion and did not know that he was jaundiced. On the basis of his complaints, a preliminary diagnosis of a peptic ulcer was, in fact, made.
The appellant offered evidence that argued against contributory negligence, if such countervailing evidence had been needed. Mr. Santoni was not accustomed to taking medication. Reactions to medication are not always clear-cut and definite. A layman is not in a position to know that ill-defined symptoms are due to a drug. Doctors themselves many times are unsure. The symptoms are sometimes very prominent; at other times they are very subtle. When the symptoms, which the evidence indicates Mr. Santoni associated with his stomach, became more pronounced, Mr. Santoni did, indeed, seek medical attention. This, furthermore, was not an isolated case where one man allegedly knew but failed to report signs and symptoms of hepatitis and died as a result. There were twelve other deaths in 1972 in Baltimore City of participants in the City’s INH prophylaxis program. There was no evidence that any of them complained of any adverse reactions. Dr. Schaerf, in fact, stated that of the six to eight deaths at the Druid Clinic, none of the patients reported any symptoms consistent with *511hepatitis. Mr. Santoni can be held accountable for what a reasonably prudent person in his position would have done.
The precise question before us, however, is not what the countervailing evidence might have been to negate contributory negligence but whether such countervailing evidence was even necessary. The burden of proof upon the issue was upon the defendants and the question is whether their best version of the facts added up to a prima facie case. The defendants argue that there was evidence from which an inference could be made that Mr. Santoni either associated or should have associated his symptoms with the medication but failed to report the adverse reactions to the clinic nurses because he was afraid to lose his job with the City. In addition to Mrs. Santoni’s testimony that her husband believed he was required to take the drug in order to work for the City, they rely upon the evidence that Mr. Santoni told Dr. Notarangelo, upon his admission to Mercy Hospital, that he was self-employed and worked at City Hall on weekends. There was also evidence that Mr. Santoni stopped taking the medication two weeks before he was admitted to the hospital.
The appellees argue that this gives rise to a possible inference of contributory negligence. The appellant argues, on the other hand, that this gives rise to the equally probable inference of due care.
It is important to state precisely the test for measuring the legal sufficiency of the evidence on the issue of contributory negligence in a case such as this. In the most routine setting, where each party has offered conflicting factual versions of the event and where each version could give rise to its own permitted inferences, the test is as stated by the Court of Appeals in Fowler v. Smith, 240 Md. 240, 246-247, 213 A.2d 549:
"Negligence is a relative term and must be decided upon the facts of each particular case. Ordinarily it is a question of fact to be determined by the jury, and before it can be determined as a matter of law that one has not been guilty of negligence, the *512truth of all the credible evidence tending to sustain the claim of negligence must be assumed and all favorable inferences of fact fairly deducible therefrom tending to establish negligence drawn. Kantor v. Ash, 215 Md. 285. Cf. Suman v. Hoffman, 221 Md. 302. And Maryland has gone almost as far as any jurisdiction that we know of in holding that meager evidence of negligence is sufficient to carry the case to the jury. The rule has been stated as requiring submission if there be any evidence, however slight, legally sufficient as tending to prove negligence, and the weight and value of such evidence will be left to the jury. Ford v. Bradford, 213 Md. 534. Cf. Bernardi v. Roedel, 225 Md. 17, 21. However, the rule as above stated does not mean, as is illustrated by the adjudicated cases, that all cases where questions of alleged negligence are involved must be submitted to a jury. The words 'legally sufficient’ have significance. They mean that a party who has the burden of proving another party guilty of negligence, cannot sustain this burden by offering a mere scintilla of evidence, amounting to no more than surmise, possibility, or conjecture that such other party has been guilty of negligence, but such evidence must be of legal probative force and evidential value. State v. Hopkins, 173 Md. 321, and cases cited; Hevell v. Balto. Transit Co., 173 Md. 327; Haddock v. Stewart, 232 Md. 139.” (Emphasis in original.)
This particular question of legal sufficiency, however, does not arise in the most ordinary setting of conflicting versions of the facts. It is rather the case that even the defendants’ best version of the facts could serve as the predicate for competing inferences, one of contributory negligence and one of due care (or at least of the absence of contributory negligence). The appellant urges upon us, in such a setting, a test of legal sufficiency as articulated in Baltimore Transit Co. v. Presherry, 233 Md. 303, 308, 196 A.2d 717: "Where evidence permits several inferences equally probable, a *513proponent must eliminate the exculpatory ones.” That statement of the test came from a significant line of Court of Appeals cases dealing with alternative inferences. It behooves us initially to examine this body of case law and to determine whether it represents, indeed, a separate test of legal sufficiency. In dealing with a situation where alternative and conflicting inferences could have arisen from a single factual predicate, the Court of Appeals stated the test for legal sufficiency in Langville v. Glen Burnie Coach Lines, Inc., 233 Md. 181, 185, 195 A.2d 717:
"At this point, we are confronted by the well-established rule of evidence that the burden mentioned above is not met by proof adduced by the plaintiff to the effect that defendant’s negligence may have caused the injuries, or even that it probably did cause them, if it also appears from plaintiffs evidence that the injuries may have resulted from some other cause for which the defendant is not responsible.”
The rule of Langville was, in turn, taken from Strasburger v. Vogel, 103 Md. 85, 63 A. 202, where the Court of Appeals held that the evidence of primary negligence in that case was not legally sufficient and should not have gone to the jury. The plaintiff there offered a version of the facts that could have given rise to an inference of negligence but could also, with equal probability, have given rise to an alternative inference that negated negligence. In dealing with such an "either/or” foundation for alternate inferences, Chief Judge McSherry said for the Court, at pp. 91-92:
"But when the plaintiff himself shows that the injury complained of must have resulted either from the negligence of the defendant or from an independent cause for the existence of which the defendant is in no way responsible, he cannot be permitted to recover until he excludes the independent cause as the efficient and proximate cause of the injury;....” (Emphasis in original.)
*514And again, at p. 93:
"It is equally obvious that where the evidence of the plaintiff is evenly balanced as to whether the cause for which a defendant may be responsible, or the cause for which he is not responsible produced the injury, the jury would have no right to disregard arbitrarily the proof which exculpated and to credit only that which inculpated — to adopt the theory which would subject the defendant to damages and to reject that which would exonerate him.”
The Langville rule was invoked by the Court of Appeals in Baltimore Transit Co. v. Presberry, supra, under circumstances where the plaintiffs own case for negligence arguably permitted the inference of negligence but also permitted, upon the same established facts, the inference that the defendant was not negligent. The Court of Appeals held that that was not a sufficient predicate to allow the case to go to the jury, saying at 233 Md. 308:
"The burden is upon a proponent to establish negligence. Where evidence permits several inferences equally probable, a proponent must eliminate the exculpatory ones. See Joffre v. Canada Dry, Inc., 222 Md. 1, 8 and cases cited. See also Langville v. Glen Burnie Coach Lines, 233 Md. 181. We think the evidence of primary negligence on the part of the Transit Co. was legally insufficient.”
In Ferguson v. Wootten, 240 Md. 186, 213 A.2d 498, the Court of Appeals dealt with the converse situation where there was a direct conflict in testimony so that the option of drawing one of several opposing inferences from one of several opposing factual predicates was properly left to the jury. The Court of Appeals there distinguished that situation from one appropriately calling for the Langville rule, saying at 240 Md. 190-191:
"The appellant argues that because of this conflict in the testimony offered by the appellee, the jury should not have been allowed to speculate on which *515version was correct. The appellant relies upon Langville v. Glen Burnie Lines, 233 Md. 181, 195 A.2d 717 (1963).
In Langville, the testimony, all offered by the plaintiffs, was that the accident was caused either by the negligence of the driver of the defendant’s bus or by a sudden and unexpected brake failure which a reasonable inspection or test of the brakes would not have revealed. We held that the lower court was correct in granting a motion for a directed verdict in favor of the defendant. The case, we found, was governed by the rule that the burden upon the plaintiff to prove affirmatively the defendant’s negligence and that it was a proximate cause of the injuries is not met by proof adduced by the plaintiff that defendant’s negligence may have caused the injuries, or even that it probably did cause them, if it also appears that the injuries may have resulted from some other cause for which the defendant is not responsible. There was no direct testimony in Langville that the defendant’s negligence, rather than the brake failure, was a proximate cause of the accident. Here, the situation is different. There was direct contradictory testimony that Bailey was negligent and that he was not. As was indicated in Baltimore Transit Co. v. Presberry, 233 Md. 303, 308, 196 A.2d 717 (1964), the evidence in Langville permitted several equally probable inferences. Here, neither the appellee’s testimony nor that of her witness, Mrs. Nock, rested on inference. There was a sharp conflict in their testimony as to whether Bailey entered on the red or the green light. This was a question of fact, not of inference, and as such was properly left to the determination of the jury. . . . The issue left to the jury was not a speculative choice between the possibilities inherent in the only testimony offered but a determination of the truth in direct, unspeculative testimony of witnesses who differed as to what had happened.”
*516In Short v. Wells, 249 Md. 491, 240 A.2d 224, the Court of Appeals also dealt with a situation where a factual dispute made submission of the case to the jury appropriate. In that case as well, however, they came to grips with the Langville rule. They mentioned it by name, quoted from it, pointed out that it "has been applied many times by this Court,” but went on to point out that "it may only be applied in a proper case and this is not such a case.” After several pages of analysis of the two arguably different measures of legal sufficiency, the Court distinguished the factual conflict in front of it in that case from the very different conflict between alternative inferences, saying at 249 Md. 496-497:
"But that is not the case here where there is more than conflicting inferences. Here there is direct as well as circumstantial evidence of negligence. It was therefore a question of fact for the jury and not mere speculation or conjecture as to what to infer ... [T]he court.. . overlooked the fact that the conflict was not in the competing inferences drawable from the circumstantial evidence but in the direct evidence. For that reason, the court should have submitted the evidence to the jury.” .. . (Emphasis in original.)
In Larsen v. Romeo, 254 Md. 220, 255 A.2d 387, on the other hand, the Court of Appeals applied the Langville rule and held that the lower court correctly refused to submit a case to the jury, where two conflicting inferences could have been drawn from the same evidence and the proponent failed to negative the exculpatory inference. In Richards v. Huntt, 255 Md. 255, 257 A.2d 412, the Court of Appeals dealt with a situation where the case should have gone to the jury, because there was a conflict in the testimony itself and not simply competing inferences possibly arising from the same facts. The court there analyzed at great length both the Langville case and Ferguson v. Wootten, supra, and stressed the two types of conflict which give rise to two different ways of measuring legal sufficiency. Most recently, the Court of Appeals in General Motors Corp. v. Lahocki, 286 Md. 714, *517410 A.2d 1039, discussed the distinct nature of the things to be measured and there held that, since a factual dispute was involved, reliance "on such cases as Strasburger v. Vogel...; Richards v. Huntt... Langville v. Glen Burnie Lines. . . and Stitzel v. Kurz, 18 Md. App. 525, 308 A.2d 430 (1973), is misplaced.”
Enough! It is time to back away from the individual cases and regain perspective. There is a danger that this and predecessor cases have unduly complicated the statement of the law, whereas our ever-present goal should be to simplify such statement. There is a danger that entangled in the underbrush of case law, we have lost sight of the larger forest. From Langville and its progeny and all of the elaboration thereon, we may be able to distill a simple and more general evidentiary statement.
The simple rule, shorn of elaboration, is that the question of whether a particular inference is permitted is one of law for the trial judge. When we are dealing with conflicting direct evidence on an issue, all that is necessary to get to the jury is some competent evidence which, if the jury in its unfettered prerogative should believe, would establish the thing to be proved. It is always for the jury to determine which of several conflicting versions of the facts to believe. By the same token, when those various versions of fact give rise to various permitted inferences, it is always for the jury to determine which inference to draw (or, in its prerogative, to draw no inference) 2. All of the language, however, about leaving to the jury the question of which inference to draw is uttered in the limited context of competing permitted inferences. When in discussing the law of directed verdicts, therefore, the case law speaks of giving the party against whom the motion is directed the benefit of the most favorable version of the facts and all inferences fairly deducible therefrom, the phrase "inferences fairly deducible therefrom” means "permitted inferences” and not mere speculative possibilities.
*518Although the jury may choose between permitted inferences (or choose not to draw any inference at all), it is for the judge in the first instance to determine, as a matter of law, which inferences are permitted. When analyzing the legal sufficiency of circumstantial evidence rather than direct evidence, we are measuring the adequacy of predicate facts, assuming them to be true, to give rise to inferred facts. The rule is that the inferred fact must follow more likely than not from the predicate fact for the jury even to be permitted the option of inferring. This is the production burden which the proponent must meet in order to establish a prima facie case and avoid the peril of a directed verdict. Even when that production burden is met, of course, the proponent must still persuade the factfinder, by the appropriate burden of persuasion, to draw the inference, but if the production burden is not met, he does not even get the opportunity to persuade.
What, therefore, appears to be a separate sub-body of case law in the Strasburger-Langville tradition for handling equally probable but alternative inferences arising from a single factual predicate actually dissolves, upon closer analysis, into nothing more than a particular and limited instance of the general principle. Deceptively, however, that body of case law developed a language all of its own without adequate reference back to the general principle. When those cases speak of the evidentiary insufficiency of showing "that the injury complained of must have resulted either from [A] or from [B]... until he excludes [B],” it is simply a way of saying that neither A nor B is a permitted inference because neither A nor B is more likely than not to be true. When those cases speak of the fatal failure to eliminate the exculpatory inferential alternative, they are simply stating the mathematical truism that if A and B begin as equally probable alternatives, A cannot become the more probable of the two and rise to the level of a permitted inference, unless and until B has either been eliminated or had the likelihood of its occurrence significantly diminished. An inferred fact that is only equally probable is not a permitted inference because, by definition, if it is only "equally probable,” it is not "more probable.”
*519Applying then this general test of legal sufficiency to the particular facts in this case, we hold that even that version of facts most favorable to the defendants was not legally sufficient to give rise to a permitted inference of contributory negligence. There was a possibility of contributory negligence. One could speculate that there might have been contributory negligence. Contributory negligence was not, however, a more likely than not conclusion to draw from the ambiguous factual predicate erected by the defendants.
From the established facts that the protocol of the U. S. Public Health Service was distributed to all health clinics in Baltimore City, was discussed at monthly meetings of the charge nurses at those clinics and was observed by an administrative nursing supervisor to be followed on several occasions, it is no more likely that questions about symptoms were specifically asked of Mr. Santoni than that they were not. The defendants would go further, however, and pile another inference upon the first inference they would have had the jury draw. Once again, however, from the fact that questions about symptoms were asked Mr. Santoni, it is no more likely that he would have deduced from the questioning the fact that a significant medical danger was involved than that he would not have so deduced. The defendants, however, would have us pile yet a third inference upon the first two. Once again, from the fact that no symptoms were recorded on Mr. Santoni’s chemoprophylaxis sheet, it is no more likely that Mr. Santoni knew of the danger and negligently failed to report the symptoms than that a laxness in the monitoring procedure produced the empty record.
From this ambiguous factual predicate, a number of mere possibilities may have followed. There was, however, no probability of contributory negligence, for there was no more reason to infer it than there was not to infer it. Since the evidence did not point toward it as a logical and probable conclusion it was not sufficient to support a legally permissible inference that Mario Santoni was negligent and that his negligence was the proximate cause of his death. The jury should not have been permitted to speculate on the *520matter. A legally permissible inference cannot be based on mere speculation or conjecture.
Having disposed of the appellant’s appeal on this issue, it is unnecessary to address the other questions raised by her.
This disposition of this issue does, however, require us to address the issues raised by the cross-appeal. The appellees/cross-appellants contend that their motions for a directed verdict on the grounds of governmental immunity should have been granted. Suit was originally brought against four doctors. All four filed motions for summary judgment on the ground that they were immune from suit because they were sued for discretionary acts performed in the course of their employment with the Baltimore City Health Department. Only the motion of Dr. Robert E. Farber, who was the Commissioner of Health of Baltimore City at the time of the isoniazid surveillance program, was granted. At the end of all the evidence, the appellees again moved for a directed verdict on the same ground. Their motions were again denied.
In their cross-appeal, the appellees/cross-appellants argue that they exercised administrative responsibilities within the Health Department which involved pure discretion for which they should be completely immune from suit. They rely on James v. Prince George’s County, Maryland, 288 Md. 315, 418 A.2d 1173, which reaffirmed that governmental immunity extends to public officials when performing discretionary, rather than ministerial, acts in furtherance of their official duties. The appellees, however, failed to establish the initial premise for that proposition — that they were indeed public officials rather than public employees. Since the question of who is a public official rather than a public employee is not always clear and involves an application of various tests (see Duncan v. Koustenis, 260 Md. 98, 271 A.2d 547), we cannot take for granted that which was not proved.
The appellees’/cross-appellants’ other contention is that the evidence of the deaths of twelve other participants in the City’s INH program was improperly admitted, placing a burden on them to disprove twelve other cases of negligence *521and prejudicing the jury. The evidence of twelve other deaths was admitted not for the purpose of showing independent acts of negligence but only for the purpose of establishing causation, existence of a dangerous situation, and knowledge of the dangerous situation on the part of the appellees/cross-appellants. It was, moreover, relevant on the issue of negating contributory negligence, as has already been fully discussed. With respect to the issues of causation and the existence of a dangerous situation, the thirteen deaths were definitely similar in circumstances. All thirteen people died of liver disease in the same year after INH ingestion as part of the City’s isoniazid program. The evidence indicated that all of the deaths were either directly, probably, or possibly related to the isoniazid. Dr. Schaerf stated specifically that of the six to eight deaths of participants at the Druid Chest Clinic, none reported any signs or symptoms consistent with hepatitis. This fact was highly relevant on the questions of 1) whether INH ingestion caused Mr. Santoni’s toxic hepatitis and was a dangerous drug, 2) whether the clinic personnel knew or should have known of its danger from the prior deaths and properly should have advised the participants of the risks and 3) whether Mr. Santoni’s death was due to some unusual dereliction or contributory negligence on his part. In the final analysis the ultimate weighing of evidentiary relevance versus its counterweight of possible prejudice is assigned to the wise discretion of the trial judge. The trial judge here found the evidence to be sufficiently relevant on a combination of issues to warrant its admission. We see no abuse of that discretion. See McCormick, Law of Evidence, § 200 (2nd Ed., 1972).
Judgment reversed in part and affirmed in part.
Case remanded for retrial on issue of damages; costs to be paid by appellees.