dissenting.
In my opinion, the evidence is insufficient as a matter of law to present a jury issue as to the defendants’ liability for breach of an implied warranty of fitness. We have long adhered to the rule that an inference must be drawn from a fact in evidence, and not from another inference. Kayh v. Commonwealth, 219 Va. 424, 427, 247 S.E.2d 696, 698 (1978), quoting from Doyle v. Commonwealth, 212 Va. 677, 678, 187 S.E.2d 201, 202 (1972). In the present *660case, however, the jury verdict approved by the trial court depends entirely upon the pyramiding of inferences.
The majority acknowledges that to create a jury issue circumstantial evidence must show an alleged result to be probable rather than merely possible. I believe that the evidence in this case falls far short of permitting any reasonable inference of probability that Doggett’s feed contained poison when it left the manufacturer and the retailer.
The established facts were that Doggett owned 66 cattle which died on January 9 from eating phorate poison marketed as “Thimet,” that the cattle were fed Pro-Blend 50, and that on January 16 a Pro-Blend 50 bag, lying with numerous other similar bags in the building where the feed was mixed, contained a tablespoon of debris. This debris, upon analysis, was found to include feed residue, rodent feces, peanut redskins, a bird feather, straw fragments, wood splinters, pieces of burlap (the bag itself was made of polypropylene, not burlap), and poison.
The first inference was that the condition of the bag one week after the death of the cattle proved its condition at the time it left the manufacturing plant and the retail store. Under the general rule, evidence of the existence of a condition subsequent to an event is some evidence of the condition’s existence at the time of the event. But the corollary rule is equally important, that evidence of a subsequent condition is incompetent to show prior existence of the condition where “the contingencies of change were too many” for the subsequent condition to have probative value. Butler v. Greenwood, 180 Va. 456, 464, 23 S.E.2d 217, 220 (1942). Thus, in Washington, &c. R. Co. v. Vaughan, 111 Va. 785, 69 S.E. 1035 (1911), we affirmed the trial court’s refusal to admit evidence that a rail car’s lights were on as it operated on its route fifteen minutes after the accident.
In the present case, the evidence was uncontradicted that Thimet has a strong and distinctive odor, but Doggett, Jr., did not detect its presence when pouring the Pro-Blend 50 into the feed mill. The Department of Agriculture inspector smelled Thimet at the feeders on January 10. The next day he smelled Thimet at the feed mill and continued to smell it as he examined several empty Pro-Blend 50 bags lying about 10 feet away, but he believed that this odor came from the Thimet that he had just smelled at the feed mill rather than from the bags. Thus, while the obvious purpose of his investigaton was to determine the source of the *661Thimet, the inspector did not then or thereafter take any of the bags for analysis. On January 16, the County Agent picked up four empty feed bags from a pile of “a dozen or more” and detected the odor of Thimet in one which he forwarded to the laboratory in Richmond. The evidence fails to show whether this bag, not received by the laboratory until January 31, had been on the Doggett premises for weeks, months, or years before the cattle died.
The inference that the condition of the contaminated feed bag one week after the accident was probative of its condition when it left the plant and the store depended upon three other inferences. First, the jury was permitted to infer that the bag was the same used by Doggett, Jr., in mixing the feed on January 7 and feeding cattle on January 9, although neither he nor anyone else ever identified the bag as the same.
Second, the jury had to infer that no intervening cause changed the condition of the bag. The probability that the bag’s condition continued unchanged from the time it was opened depends “on the nature of the specific fact and the circumstances of [the] case.” Butler, 180 Va. at 464, 23 S.E.2d at 220. The evidence in the present case negated the reasonableness of any inference that the bag’s condition remained unchanged after opening. When discovered a full week after the accident, the bag was one of many bags lying on the floor, and its contents, except the feed residue, were inconsistent with the usual contents of Pro-Blend 50 bags. Indeed, no other bags of Pro-Blend were found to contain any Thimet, although Doggett submitted a sample of feed from a bag that he reported had been used by his son in the fatal feeding operation.
Third, the jury had to infer that the bag contained at some point sufficient poison to kill 66 cattle. Doggett’s expert testified that the poison concentration in the tablespoon of debris found in the bag was 1.46%. Then assuming an even mixture of 1.46% in a 100-pound bag, he calculated that the full bag would have contained a lethal concentration of approximately one and one-half pounds of poison. (The evidence of Southern States was that it was impossible in the manufacturing process for Thimet to have been packed in a Pro-Blend 50 bag, and in any event, impossible for one and one-half pounds of any foreign substance to have been packed in one bag without a similar quantity being injected into 39 other bags packed automatically at the same time).
*662In my opinion, this leap from inference to inference to inference to inference, while impressive as a display of mental agility, cannot be justified. The jury should not have been permitted to reach a verdict based upon speculation, sympathy, and quadruple inferences.
Moreover, while it may be arguable that Instruction 9A, proffered by Southern States and refused by the trial court, was unnecessary, I do not agree that it was misleading. It is true, as the majority states, that there was no doubt that Thimet in the feed killed the cattle. But that does not justify refusing an instruction expressing the Southern States theory that there were many ways, for some of which Southern States was not responsible, that the Thimet could have been introduced into the feed. For an obvious example, the Thimet could have been poured from a Thimet container into the Doggett feed mill accidentally by one of Doggett’s employees. There was evidence of the presence on the premises of two Thimet containers, one an empty box near the feed mill, and the other an unopened bag in another building. Doggett admitted that he had used Thimet to control rodents within four to six months prior to the accident.
I would reverse the judgment of the trial court and enter judgment for Southern States and Farmers Service.
HARRISON, R.J., joins in dissent.