LAY, Chief Judge.
This federal tort claim suit arises out of the February 13, 1980, crash of a twin engine airplane near Hot Springs, South Dakota. The airplane, owned and piloted by Charles M. Largent, crashed shortly after takeoff from the Hot Springs airport with three passengers aboard, one of whom was Louis Altringer, a business associate of Largent. All aboard the plane were killed. The estates of Largent and Altringer brought this action against the United States, alleging that the government negligently failed to warn Largent of adverse weather conditions that led to structural icing causing the plane to crash.
The district court1 found that the government was negligent but that icing did not cause the crash and entered judgment for the United States. The court also found that Largent was contributorily negligent and that his negligence was more than slight, barring his claim under the South Dakota comparative negligence statute. See S.D. Codified Laws Ann. § 20-9-2 (1979). It further found that Altringer was engaged in a joint enterprise with Largent in the operation of the aircraft, thus barring Altringer’s claim through the doctrine of imputed contributory negligence. The estates of Largent and Altringer appeal. Facts
Charles Largent was the president, manager, and a shareholder of Fall River Feedlots, Inc., a corporation headquartered in Hot Springs, South Dakota. Louis Altringer was the assistant manager and a share[1332]*1332holder of Fall River Feedlots. On February 13, 1980, Largent, Altringer, and two officers of the First National Bank of Hot Springs planned to fly in Largent’s twin engine Beech Baron aircraft from Hot Springs to Modesto, California, to inspect some cattle.
Government Negligence
At 4:16 a.m., Mountain Standard Time, from his home in Hot Springs, Largent telephoned the Flight Service Station (FSS) in Rapid City, South Dakota, and spoke with FSS Specialist Charles Shields. An FSS is a Federal Aviation Administration air traffic facility that provides services to airmen such as weather briefings, receiving and processing flight plans, and communicating with airborne aircraft. Because Hot Springs had no FSS facility, it was the practice of Hot Springs pilots to telephone the Rapid City FSS to obtain weather briefings and flight clearances and to file flight plans.
Largent identified himself to Shields as Baron 8592R, informed Shields that he would be departing from Hot Springs at 6:00 a.m. destination Modesto via Casper, Rock Springs, Salt Lake City, and Reno, and asked for a weather briefing. Shields gave Largent general weather information such as cloud densities and ceilings at all points along Largent’s intended route. Largent also asked for and was provided with winds aloft information. Largent then filed his flight plan (it is customary to do this over the phone), Hot Springs direct to Casper direct to Rock Springs, and requested an altitude of 8,000 feet mean sea level (MSL). At the end of the fifteen-minute conversation, Shields informed Largent that Rapid City had been IFR (Instrument Flight Rules, i.e., overcast) all night, and the cloud ceiling had dropped to about 700 feet within the last three hours. Largent informed Shields he would call again just before takeoff to get a clearance void.
Shields did not inform Largent of an area forecast that warned of icing over South Dakota and eastern Wyoming. Shields testified that this forecast had been issued at 5:40 p.m. on February 12 and would expire at 6:00 a.m. on the thirteenth; thus, he believed it was no longer accurate. Shields knew that Largent would be calling back at 6:00 a.m., just before departure, and he thought he would have a new forecast by then. If icing was still forecast, he could inform Largent of these conditions at that time.2
However, when Largent again called Rapid City FSS just before 6:07 a.m., he did not speak with Shields but with specialist Paul Kasen, who had come on duty just two minutes before. Kasen did not know that Shields had postponed warning Largent of the icing forecast. Largent requested his flight clearance and after receiving it, asked Kasen, “What have you got up there?” Kasen informed Largent that Rapid City had light snow showers with broken clouds at 700 feet, overcast at 3,000 feet, and visibility of three miles. Largent also asked for conditions in Casper and Rock Springs. Kasen gave a statement to investigators that he observed in the FSS pre-flight briefing log that Shields earlier had given Largent a complete weather briefing and that Kasen therefore only provided the information specifically requested.
The district court found that Shields was negligent, but that Kasen was not. The court noted that the FSS handbook does not permit specialists to amend forecasts, which is essentially what Shields did by disregarding the icing forecast. The court also found that Kasen was not negligent because Largent informed Kasen that he was calling for a takeoff clearance and did not mention a weather briefing.
The record reveals that at 5:40 on the morning of the flight, the National Weather Service issued a Kansas City area forecast, which covered Largent’s route of flight. The forecast was effective from 6:00 a.m. to midnight of February 13, and called for moderate mixed icing in the clouds over southwestern South Dakota [1333]*1333and generally deteriorating weather conditions through the morning hours. It is unclear exactly when this forecast was received at the Rapid City FSS, but Shields testified that it should have been received at 6:00 a.m. Had it not been received at this time, he stated, it would have come in on a teletype scan at 6:40 a.m. The government’s expert meteorologist testified that at this time, the National Weather Service was converting to a new communications system, occasionally causing delayed transmissions. There is no evidence of whether Rapid City FSS received the new forecast at 6:00 a.m. It is not disputed that Kasen did not give Largent the new icing forecast.
The district court declined to speculate as to whether Kasen had received the updated forecast by the time Largent spoke to him. The court found that Kasen was under no duty to give Largent another weather briefing under the guidelines in the FSS handbook because “Largent was calling for his clearance, and not a weather briefing.” We deem this finding clearly erroneous. The facts are undisputed that after receiving his clearance, Largent specifically inquired into weather conditions along his route of flight by asking Kasen, “What have you got up there?” Kasen understood this question to be weather-related as indicated by his answer, “Oh we’ve got the sky partially obscured estimated seven hundred broken three thousand overcast three miles. Light snow showers.” Largent then asked for conditions reported in Casper and Rock Springs and asked Kasen to recheck winds aloft at 9000 feet. By inquiring into these weather conditions, Largent went beyond a mere request for clearance, as the district court characterizes the conversation with Kasen, and thus imposed an additional duty upon Kasen to provide him with pertinent weather conditions relative to his flight, especially adverse and potentially dangerous conditions such as icing.
Whether Kasen had received the updated forecast from Kansas City is immaterial. He knew by looking at the briefing log that if Shields did convey to Largent the icing forecast, that forecast was ten hours old at the time Largent would have received it. He also knew that Kansas City would have issued a new forecast twenty-five minutes before his conversation with Largent. Kasen knew the old forecast was to expire at 6:00 a.m. that morning. Had Kasen been exercising due care, Largent’s request for further weather information should have prompted Kasen to at least report in full the old forecast including icing warnings and to ascertain whether the old forecast which expired at 6:00 a.m. had been updated. Kasen must have known that the weather conditions had been deteriorating in the area. The Sixth Circuit has held that “[s]ince the FAA has undertaken to advise requesting pilots of weather conditions, thus engendering reliance on facilities such as the Indianapolis FSS, it is under a duty to see that the information which it furnishes is accurate and complete.” Pierce v. United States, 679 F.2d 617, 621 (6th Cir.1982). Complete information means current information. Even had the updated forecast not been received in Rapid City, a simple telephone call by Kasen to the weather service would have provided him with the updated forecast. We hold that the district court’s finding that Kasen was not negligent is clearly erroneous. Both Shields and Kasen as flight service specialists were grossly negligent in failing to provide Largent with accurate weather information. The evidence is undisputed that the government agents deprived Largent from having the correct weather information so as to allow him to make the critical judgment as to whether to fly the plane into icing conditions. If the proper forecast of current icing conditions had been given to Largent it is beyond speculation that he would have chosen not to take off in an aircraft that was not equipped with de-icer equipment.
Largent’s Negligence
The government contends, and the district court alternatively found, that even if icing could be found to have caused this crash, Largent as pilot of the aircraft was guilty of more than slight negligence when compared with the negligence of the [1334]*1334government. Defendant’s expert witness, Bernard Coogan, testified that one of the most basic lessons given to pilots learning to fly IFR is to recognize and be alert to conditions under which icing is likely to occur. Two conditions must exist for icing to occur: (1) the aircraft must be flying through visible moisture such as rain or clouds; and (2) the temperature must be below freezing. Given the conditions at Hot Springs that morning — low overcast, temperature in the twenties Farenheit, and very light snow — Coogan stated that although he would expect to be told of an icing forecast by the FSS specialist, if no mention was made of icing, he would have asked if icing was forecast. Even without such a forecast or pilot reports of icing, Coogan asserted that the phenomenon can still occur and a careful pilot would plan for it.
Coogan did not believe Largent asked Shields for the information necessary to make an informed decision as to whether he faced a “go or no go” situation. For instance, Largent did not ask Shields for “tops,” maximum cloud heights, which are important to know if one encounters icing. According to Coogan, if a pilot flies into an icing situation and he knows the tops, he can sometimes continue his climb and get out of the danger zone, but only if he knows how much higher he needs to go. Largent did not have this information. Also, Largent did not ask for temperatures aloft when Shields was reporting the winds aloft. Coogan stated that this information is helpful because the colder it is, the less the chance of icing. By not requesting certain pertinent information that a pilot of his experience should have recognized as necessary, Coogan testified, Largent did not leave himself an out should he have encountered icing. James Strang, Largent’s flight instructor, corroborated this opinion. He stated that Largent had the ability and experience to recognize that he would be likely to encounter icing on this flight. Strang drove past the airport later that morning on his way to the crash site and stated that he would not have flown into that weather.
Even George Rhodes, plaintiffs’ own expert pilot, testified that when flying in winter conditions, a prudent pilot would ask for a report on icing if none were given during the weather briefing. Rhodes stated that icing can occur at any time and when the conditions are right, a pilot should always leave himself an out. In doing so, Rhodes said, it would “be nice” to know the tops and the temperatures aloft, information which Largent did not request.
Another defense witness, Willis Bublitz, was a pilot for the forest service who encountered icing near Hot Springs later that morning. Bublitz said that he anticipated icing that day just by observing the weather conditions. He stated that before taking off in the winter, he always called the weather service and asked for ice reports and top reports. The defendant’s expert Allen Pearson opined that the conditions necessary to produce moderate icing were not present. He based this conclusion largely upon radar reports from Rapid City and Alliance which in his opinion, showed only “rime” icing conditions existing in those areas. As the defendant indicated, however, the only means for direct observation of icing are pilot reports from aircraft in flight. Although there was conflicting testimony respecting the interpretation of the weather data for the Hot Springs area, the evidence is uncontroverted that two separate pilots flying in the forecast area reported icing on their aircraft. The defendant’s evidence ignores that it would take only a small degree of airplane icing on the wings of Largent’s plane to affect the lifting power of the aircraft. The evidence must also be evaluated along with the current NWS forecast concerning Largent’s actual route that specifically warned of icing conditions which the government negligently failed to disclose to Largent, as well as the actual icing experience of the ground witnesses. Even the trial judge found that icing conditions existed at the time and site of the crash.3
[1335]*1335FAA regulations mandate that the pilot in command of an aircraft is directly responsible for, and is the final authority as to, the operation of that aircraft. 14 C.F.R. 91.3(a) (1986). Largent’s Beech Baron was not equipped with deicing equipment other than a pilot heater, which keeps the airspeed indicator functioning, and a propeller anti-ice fluid system. Thus, he had no apparatus, such as inflatable boots, to remove ice from the wings. He should have been especially alert to the possibility of icing and should have been prepared to deal with it. The evidence supports the trial court’s finding that Shields was negligent, and although arguably Shields’ description of “good” weather could have lulled Largent into such belief, we agree with the district court that Largent was guilty of some negligence. He was in the immediate area and could observe the local ground conditions. Notwithstanding the government’s negligence, Largent failed to ask either Shields or Kasen for relevant weather information crucial to the safe conduct of this flight. We find substantial evidence exists that supports the trial court’s finding that Largent was guilty of some negligence and under the circumstances we cannot say that finding is clearly erroneous.
South Dakota Codified Laws Annotated § 20-9-2 (1979) provides:
In all actions brought to recover damages for injuries to a person or to his property caused by the negligence of another, the fact that the plaintiff may have been guilty of contributory negligence shall not bar a recovery when the contributory negligence of the plaintiff was slight in comparison with the negligence of the defendant, but in such case, the damages shall be reduced in proportion to the amount of plaintiff’s contributory negligence.
The district court found that Largent’s negligence was more than slight in comparison with the negligence of the government. “Slight” negligence as used in the statute means “small in quantum in comparison with the negligence of the defendant,” rather than the common standard of the reasonably prudent person. Crabb v. Wade, 84 S.D. 93, 97-98, 167 N.W.2d 546, 549 (1969); see also Lovell v. Oahe Elec. Coop., 382 N.W.2d 396, 399 (S.D.1986). What constitutes slight or small varies with the facts and circumstances of each case. Urban v. Wait’s Supermarket, Inc., 294 N.W.2d 793, 796 (S.D.1980); Crabb, 84 S.D. at 98, 167 N.W.2d at 549. We have previously held in applying South Dakota law:
[Tjhree factors may properly be considered in appraising the quality of a plaintiff’s negligence: the precautions he took for his own safety; the extent to which he should have comprehended the risk as a result of warnings, experience, or other factors; and the foreseeability of injury as a consequence of his conduct.
[1336]*1336Associated Engineers, Inc. v. Job, 370 F.2d 633, 641 (8th Cir.1966); see also Lovell, 382 N.W.2d at 399. Taking these factors into consideration, we believe that substantial evidence exists such that the district court could find Largent’s negligence to be more than slight in comparison with the negligence of Shields.
However, the difficulty with the trial court’s finding is that it must compare Largent’s negligence with the total negligence of the government.4 We have held that the court erred in its analysis that Kasen was not negligent. Under the circumstances, the trial court must compare Largent’s conduct with the overall conduct of both Shields and Kasen. We think the record is clear that Kasen owed a duty to provide Largent with the most recent weather forecast. Thus, although some evidence exists to support the finding that Largent may have been negligent, whether such negligence was more than slight when compared with the overall negligent conduct of the government is a question that should be passed upon by the trial court.
Kasen and Shields were acting in concert on behalf of the government. Shields had earlier told Largent that between Rapid City and Salt Lake the weather was “all good.” Largent did not know that Shields intended to tell Largent of any update on the forecast two hours later. The government cannot rely on the independent fact of Shields’ “state of mind”— that is, the intent to update the forecast— and Kasen’s alleged ignorance of Shields’ intent. As far as evaluating the government’s conduct, the conduct of Shields and Kasen must be looked upon as one continuous act by a single entity. In other words, the government agents had a duty to tell Largent, when he called back, the current forecast or if none was available to tell Largent that the earlier forecast was incomplete and must be amended to include icing conditions. If Shields withheld weather data (icing conditions) from the first report and intended to update Largent at the time that Largent was scheduled to call back, Kasen was duty bound as Shields’ co-employee to tell Largent the correct information.
Under the circumstances, we must remand to the trial court to reevaluate the evidence concerning Largent’s negligence in terms of the comparison of both Shields’ and Kasen’s conduct. In doing so we emphasize that Shields’ conduct, although negligent, was somewhat justified by his assumption that he could amend any changes in the forecast when Largent called in at 6:00 a.m. To urge that Kasen was not negligent because he relied on Shields’ incomplete briefing placed Largent in a “Catch-22” situation.
As we will discuss, we find the trial court erred as a matter of law in holding that Altringer was barred by reason of the existence of a joint enterprise. However, before discussing this issue, it is necessary to pass on the trial court’s finding that the negligence of the government was not a proximate cause of the crash.
Causation
Plaintiffs assert that this crash was caused by structural icing on Largent’s airplane. The government claims the evidence shows that Largent became spatially disoriented and lost control of the airplane when he flew into the overcast because he was not qualified to fly into IFR conditions. The district court held that the plaintiffs failed to prove that icing was the proximate cause of the crash. It further observed: “This Court declines to speculate as to the specific cause of the crash. However, the Defendant’s theory that Mr. Largent lost control of the aircraft due to spatial orientation is a more plausible explanation especially in light of the fact that mechanical failure and icing have been ruled out.”
[1337]*1337We may reverse the trial court’s finding when although there is evidence to support it, we are left with the definite and firm conviction that a mistake has been made. Anderson v. Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) (citing United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 541, 92 L.Ed. 746 (1948)); see also Taylor v. Pre-Fab Transit Co., 616 F.2d 374, 375 (8th Cir.1980). Having reviewed all of the evidence presented, we possess a firm and definite conviction that the trial court’s findings on proximate cause are both devoid of evidentiary support and mistaken.
Spatial Disorientation
Contrary to the district court’s speculation, overwhelming evidence showed that Largent did not become spatially disoriented. According to the evidence presented, a pilot may experience spatial disorientation when he flies into a cloud bank and can no longer see the horizon. The evidence shows that in the clouds, he loses all visual references and may rely completely upon the balance perceptions registered in the semi-circular canals of his inner ear. Bernard Coogan testified that the semi-circular canals react to a change in the attitude of one’s head, but the inner ear does not perceive very slight and gradual changes in attitude. Gravity, the thrust of the airplane, centrifugal force, turbulance, or the pilot’s head movements, Coogan testified, can aggravate these false perceptions to such an extent that the pilot literally loses sense of where the ground is. Thus, before a pilot is permitted to fly solo into IFR conditions, he must become proficient in flying solely on instruments. Coogan stated that by learning to ignore his false balance perceptions and trusting the instrument readings, a pilot can fly in IFR conditions without incident.
The Beech Baron Pilot’s Operating Manual explains:
The result of vertigo [spatial disorientation] is loss of control of the airplane. If the loss of control is sustained it will result in an excessive speed accident. Excessive speed accidents occur in one of two manners, either as an inflight airframe separation or as a high speed ground impact; and they are fatal accidents in either case. All airplanes are subject to this form of accident.
The record reveals that inflight airframe separation occurs when a pilot experiencing spatial disorientation recognizes that his perceptions are false and upon reading the flight instruments, overcorrects. This causes the various turn, pitch, and altitude controls to exert so much pressure on the airframe that the airplane literally breaks apart in mid-flight. There is no evidence or any allegations that Baron 8592R underwent this type of stress. The alternative was that an excessive speed ground impact occurred. “Excessive speed” is defined in the Beech Baron manual as an airspeed “greatly in excess of two operating limitations which are specified in the manuals: Maximum manuevering speed and the ‘red line’ or ‘never exceed’ speed.” The manual states that the red line or never exceed speed for the Beechcraft Baron B55 is 233 knots or 257 miles per hour. The maximum maneuvering speed is 156 knots or 180 miles per hour.
The evidence in the present case shows that the airplane did not crash at an excessive speed as defined in the manual. Plaintiffs’ expert, George Rhodes, opined that the plane impacted the ground at a speed between 90 and 110 miles per hour in a slightly nose down attitude. He also stated that it crashed onto hard, frozen, clear ground on a ten degree downslope, but the record shows that the plane came to rest only twenty-two feet past its point of impact. The only inference available is that the plane crashed at a relatively slow speed. Defendant’s own expert accident reconstructionist testified that the airplane could not have been traveling at more than sixty miles per hour upon impact.
Additionally, if one were to accept spatial disorientation as the cause of this crash, one would have expected the plane to show signs of having been out of control at impact. There was no such evidence here. The evidence showed that when the plane crashed, the wings were in approximately a [1338]*1338thirty-degree left bank attitude, which was essentially level with the ground at the point of impact. The evidence also showed that the plane clipped a small pine tree at a point fifteen feet above ground, approximately 100 yards from the point of impact. In other words, it flew a fair distance past this tree. Rhodes testified that this was “a rather good crash landing.” He stated that the plane was not out of control and that the pilot did “a commendable job in doing what he should have done.” Defendant’s expert, William Allen, did testify that this was an uncontrolled crash, but in an earlier deposition, he testified that Largent was trying to make a crash landing.
Allen also testified that when Largent flew into the clouds, he turned in the wrong direction, which the government asserts supports its spatial disorientation theory. The evidence showed that the plane took off in a northerly direction and then turned to the left, heading west. Allen stated he would not have anticipated that Largent would turn left to fly into the Black Hills at such a low altitude in IFR conditions. He also noted that Largent’s heading selector was set to 165 degrees and that Largent had the 270 degree course to Casper tuned in on his radio, which indicated to Allen that Largent would have made a right turn after takeoff to 165 degrees and climbed on that course until he was safely above the terrain to the west or until he intercepted the Casper VOR.5 However, Allen’s testimony does not take into account that Largent filed his flight plan out of Rapid City. An examination of the maps in evidence shows that had Largent intended to fly south until he intercepted the 270 degree VOR heading out of Cheyenne, he would essentially be flying to Chadron before heading west. Chadron also has an FSS, and had this been Largent’s intended route of flight, it is equally plausible that he would have called Chadron to obtain his weather briefing and file his flight plan.
The evidence shows that Largent did in fact intend to make a left turn and fly in a northwesterly direction. Several factors in the record reinforce this conclusion. First, Largent was cleared by Denver Air Traffic Control (ATC) to fly at an altitude of 10,000 feet. Exhibit twenty-three, which is the Cheyenne sectional aeronautical chart in effect at the time of the crash, shows that although Casper is almost due west of Hot Springs, a pilot without a special clearance flying at an altitude of less than 12,000 feet must fly around the Edgemont Military Operations Area which lies between Hot Springs and Casper. To do this, one must fly either straight south toward Chadron and then west to Casper or northwest toward Newcastle and then west. Our calculations show that the Chadron route is approximately 172 miles while the Newcastle route is only 156 miles.
Second, an examination of the sectional map shows that flying northwest, the pilot would encounter Low Altitude Federal Airway V26 that connects Rapid City and Casper. Airway V26 is a controlled airway. The Airman’s Information Manual received in evidence indicates that certain requirements exist for operating in a controlled airway, including flight only at the altitude assigned by the ATC. Kasen had informed Largent before takeoff that Denver ATC had assigned him an altitude of 10,000 feet. This assignment suggests that Largent intended to fly in a controlled airway, and the southerly route suggested by Allen does not intercept such an airway.
Finally, Judy Kaufman, a friend of Largent’s, testified that Largent had told her he flies over her house when he flies west. Our study of the maps shows that Kaufman’s house is located approximately seven miles northwest of the Hot Springs Airport on a course one would expect a pilot flying from Hot Springs to get onto V26. Thus, the evidence shows that in all probability, Largent intended to turn left shortly after takeoff.
[1339]*1339Largent departed from the Hot Springs Airport at 6:27 a.m. Kaufman testified that at 6:30 or 6:35, she heard an aircraft fly over her house from the north towards the southeast at what she described as a “very low” altitude. According to the National Transportation Safety Board, Largent’s plane crashed at approximately 6:38, three miles from and facing in the direction of the airport. The crash site is directly between Kaufman’s house and the airport. The flight path suggested by this evidence is indicative of a pilot who is in trouble and is trying to return to his departure point. A disoriented pilot in all likelihood would not have taken such a logical route. All of this evidence in addition to that showing that the plane crashed at a speed just above the stalling speed,6 in essentially a wings-level position and slightly nose down, demonstrates that the pilot flying this aircraft exercised a fair amount of control over its flight. On the basis of these evidentiary facts, we conclude, as did the trial court, that the record does not support a finding that the crash was caused by spacial disorientation.7
Icing
As the trial court recognized, the burden was on the plaintiffs to prove that the negligence of the government in failing to warn of icing conditions was a proximate cause of the crash. The evidence shows that structural icing is a potentially dangerous condition that can occur anytime there is visible moisture in the air and the temperature is below the freezing point. As Rhodes explained, icing not only adds weight to the aircraft, more significantly, it alters the aerodynamic performance of the affected surfaces on the aircraft thus reducing airspeed and hindering the lift capacity of the wings. Rhodes stated that icing can progress to the point where the airplane is no longer capable of flying. The Airman’s Information Manual states that as little as one-half inch of ice can reduce the lift capacity of some aircraft by as much as fifty percent and increase frictional drag by an equal amount. Icing increases an aircraft’s stalling speed and decreases its cruising speed. Rhodes testified that even a thin coat of frost has been known to prevent some planes from being able to take off. Rhodes also said that ice can form on a windshield and impair the pilot’s vision, which can be very troublesome psychologically to the pilot.
The conditions near Hot Springs at 6:30 a.m. on February 13, 1980, were certainly ripe for icing. The parties stipulated that the Kansas City area forecast known to Shields and in effect until 6:00 a.m. on February 13, 1980, contained a flight precaution for South Dakota and Wyoming for icing, ceilings below 1,000 feet, and visibilities of less than three miles. The National Weather Service forecast called for light to moderate mixed icing in clouds and precipitation freezing at the surface. The area forecast effective at 6:00 a.m. on February 13 contained a flight precaution for brief moderate, mixed icing until 9:00 a.m. over southwestern South Dakota. The record also reflects that weather conditions generally deteriorated throughout the early morning hours.
There was also testimony by several witnesses of icing on the ground. Gerald Bruce testified that he left for work from his home, located about four miles northwest of the crash site, shortly after 6:00 a.m. on that morning. He had to scrape approximately one-eighth inch of ice off of his windshield before he left for work. As he drove toward Hot Springs, he had to drive downhill about two and one-half miles, then back uphill again. As he ascended, at approximately 6:20, he entered the clouds or a fog bank at 3900 feet MSL. Over the course of the one mile that he drove through this cloud bank, he had to [1340]*1340stop three times to scrape the ice off of his windshield even though his car had warmed up and his defroster was on. Each time the windshield iced over, he stated, the accumulation of ice was sudden.
Judy Kaufman, who also lives in the vicinity of the crash site, testified that when she heard Largent's airplane fly over her house that morning, she stepped outside and noticed ice crystals on her deck. At 7:45, she noticed an eighth to a quarter-inch of ice on her windshield as she got ready to go to work.
Jim Geddis was driving his truck south on Highway 79, just north of the Hot Springs Airport, when he encountered a fog bank. He stated that at that time, he experienced a sudden coating of clear ice an eighth to a quarter-inch thick over the entire truck. As he was standing outside looking at the truck, he heard a boom that he later came to believe was the sound of the plane crashing.
Joe Garnet also heard Largent’s plane fly over his home that morning. He stated that he heard the plane’s engines rev and he immediately sensed that something was wrong. He concluded that the plane was too low and the engines were rough. Fifteen seconds later his wife told him that she heard the plane crash. Garnet telephoned the sheriff to notify him that a plane had gone down nearby and set out to find the wreckage. Garnet drove part of the way, then hiked over a ridge. As he hiked up the ridge, he noticed that the temperature got warmer. When he got to the top of the ridge, he surmised that the pilot of the aircraft was probably trying to fly over the ridge to get on the down slope on the opposite side where, heading into the fifteen to twenty mile per hour wind present that morning, he could pick up some airspeed and put the airplane down on the more level ground to the east. Garnet testified that there would have been no possible hope of ditching or belly landing the plane in the valley where his house was located without bad results.
Garnet found the plane and said he could find no signs of life among any of the passengers. At the crash site, he found precipitation in the form of water droplets. It was not frozen. He estimated the cloud ceiling at the crash site to be 150 to 200 feet. Garnet spent about ten minutes at the crash site and briefly inspected the plane. He stated that he ran his fingers over the leading curve of the left wing and felt moisture, but no ice. He saw no ice around the aircraft.
Forty-five minutes later, Garnet returned to the crash site with Sheriff Leo Bray. Bray stated he found twenty to forty thousandths of an inch of ice on the leading edges of the wings of the plane, but he saw no ice on the ground. Sometime later, Largent’s instructor, Jim Strang, arrived from Chadron. Strang pointed out to Bray dust trails on the air intake of an engine on the plane, which indicated where ice had been. Bray and Strang apparently did not search the entire plane; Strang merely pointed this out to Bray as an observation. Strang was not an investigator; he was merely at the site to look the wreckage over since he had recently done mechanical work on the aircraft. Bray stated that he did not do an extensive examination for evidence of icing.
National Transportation Safety Board investigator Richard Brandiger examined the wreckage at about 10:00 a.m. that day. He found an eighth of an inch of ice on the leading edge of both wings and the horizontal stabilizers. He testified that the ice ran back three inches from the leading edge. He found no ice anywhere else on the plane nor any ice on the ground nearby. Other investigators later found pine cones, needles, and bark inside the wreckage and wondered if the plane had hit a tree before crashing. A subsequent search of the crash area revealed damage to a small pine tree at the top of the ridge that Largent had crashed on and a broken piece of plexiglass that apparently resembled a landing light cover. Investigators concluded that the plane brushed the tree with its left wing about fifteen feet above ground and crashed approximately 100 yards further down the hill.
The trial court based its conclusion that ice did not cause this crash on the fact that Garnet found no ice on the aircraft or near [1341]*1341the plane when he found the wreckage, The evidence shows, however, that Garnet made only a cursory inspection of the wreckage and did not search the point of impact nor the area of the pine tree struck by the plane. This is understandable since determining the cause of the wreck was likely not Garnet’s primary thought at the time. Given the force of the impact that occurred (the evidence showed that the plane was traveling approximately 100 miles per hour but skidded or bounced only twenty-two feet beyond impact; pictures revealed that virtually everything in the cockpit was forced out the front of the aircraft), it is not only conceivable but likely that had structural icing occurred, most of the ice would have been jarred loose from the airplane and smashed upon contact with the ground. It is also likely that the pieces of ice left over melted immediately or at least shortly after impact. Garnet testified to a temperature inversion as he climbed the ridge and stated that he found liquid, not frozen, precipitation at the crash site. Sheriff Bray did not begin to examine the crash site until nearly two hours after the actual crash which would be enough time for the temperature inversion to lift or dissipate and would explain the ice Bray found on the wings.
Although the evidence showed that the Beech Baron should have been capable of flight with an eighth to a quarter inch of ice on the wings, it is clear from Rhode’s testimony that even that amount could significantly alter the performance of the plane.8 Given the suddenness with which the icing occurred that day (both Geddis and Bruce stated that icing of their vehicles was virtually instantaneous), the fact that the airplane was fully loaded if not slightly over its weight limitations,9 and the fact that Largent most probably was not expecting icing difficulties, the only plausible conclusion to be reached is that icing brought this aircraft down.
The government attempts to refute this reasoning in several ways. First, it offers the testimony of Bublitz, the pilot for the United States Forestry Service who encountered light icing later that morning in a similar airplane. It argues that Bublitz, who also had three full-grown male passengers aboard, picked up an eighth to a quarter inch of airplane icing and had no difficulty continuing his ascent. The government overlooks several other factors. First, Bublitz’ plane differed from Largent’s in two important respects: it had more horsepower and it was equipped with inflatable deicing boots that would prevent buildups of substantial amounts of ice on the plane wings. Thus, when he encountered the icing, Bublitz knew that if the plane experienced too great of a build-up, he could activate the boots and rectify the problem. Largent, on the other hand, had no icing equipment. His only prudent alternative upon encountering the icing would be to try to drop down under the clouds and turn around and head back to the Hot Springs airport. This is exactly what the evidence shows he attempted to do.
Second, the government argues that the evidence is insufficient to prove that Largent encountered an ice storm. It notes that the roads on which Kaufman, Geddis, and Bruce drove were not icy. It points out that Kaufman testified she did not experience icing once she started driving and the radar data from that area did not indi[1342]*1342cate substantial amounts of moisture in the air. Furthermore, the government argues, Garnet found no ice on his truck that morning and Geddis could not identify where he encountered the icing on his truck. This argument too, however, overlooks other significant evidence. All of the ground witnesses who experienced icing were at or above the 3900 foot MSL level at the time they experienced the icing. Bruce experienced icing as he ascended a ridge at the 3900 foot mark. The elevation of Bruce’s house where he first found his car covered with ice was 4140 feet MSL. Kaufman lived at 3968 feet, where she too found her car covered with ice. Garnet, however, who found no ice on his truck that morning, lived only at the 3223 foot level.
The elevation at the Hot Springs airport is 3143 feet. With a 700 foot ceiling, it is reasonable to infer that Largent encountered icing shortly after entering the clouds, at approximately 3900 feet. In addition, the evidence showed that the potential for hazardous icing is greater on the windward side of a mountain range. With a twenty mile per hour northeasterly wind, and conditions admittedly ripe for light icing at-the very least, it is clear that Largent was flying directly into potentially hazardous icing conditions as he climbed to 3900 feet and turned directly into the windward side of the Black Hills. It was somewhere near here, on Highway 79, that Geddis experienced the instantaneous accumulation of a quarter inch of ice on his truck.
Third, the government contends that had Largent been surprised by instantaneous icing, he would have immediately notified Denver ATC by radio. However, plaintiffs’ expert, Rhodes, who has gone down twice in airplanes, testified that the pilot’s first priority in an emergency situation is to fly the aircraft. If he is encountering trouble, he is probably going to be terrified, as are his passengers. Not only did Largent have to try to land safely, Rhodes surmised, he may have had to deal with three panicking passengers. In addition, Rhodes testified that often the first part of an airplane to ice over is the radio antenna. This, he stated, breaks radio communications.
The parties presented the court with only two possible alternatives: structural icing or spatial disorientation by the pilot. Based upon the evidence presented, we conclude the only permissible inference that can be drawn from all of the evidence is that airplane icing was a substantial factor in the crash. In this regard, spatial disorientation as a causal factor is only speculative. We possess a firm and definite conviction that the overwhelming evidence demonstrates that.
We thus conclude it was clearly erroneous for the trial court to find that the government’s negligence was not a substantial factor in the cause of the airplane crash.10
[1343]*1343Joint Enterprise
The government does not claim that the passenger, Altringer, was guilty of any active negligence. The government urges that Altringer is barred from recovery by reason of the imputed negligence of Largent, the pilot, under the doctrine of joint enterprise. The trial court so held.
Although this court normally gives deference to the trial judge’s determination of state law, we cannot do so here. The court’s analysis relies on the few cases in South Dakota involving imputed negligence under a joint enterprise theory in the driving of an automobile. The trial court found as follows:
Even if icing was the proximate cause of the accident, Mr. Altringer would be barred from recovery because Mr. Largent and Mr. Altringer were involved in a joint venture. The essential elements for a joint venture are: (1) an agreement, express or implied, among the members of the group; (2) a common purpose to be carried out by the group; (3) a community of pecuniary interest in that purpose among the members; and (4) an equal right to a voice in the direction and control of the enterprise, which gives an equal right of control. Fredrickson v. Kluever [82 S.D. 579], 152 N.W.2d 346, 348 (S.D.1967).
There was an agreement that the four persons would travel together in Mr. Largent's plane to inspect the cattle. As both Mr. Largent and Mr. Altringer were shareholders in Fall River Feedlots, they had a pecuniary interest in the cattle that they had purchased. Mr. Largent, as manager, and Mr. Altringer, as assistant manager, both had an equal right to a voice in the direction and control of the cattle inspection. Therefore, a joint venture existed. See Scheuring v. Northern States Power Company [67 S.D. 484], 294 N.W. 175 (SD 1940) (Being parties to the same enterprise, they are assumed to have common control and possession of the machine). Equal right of control is not equivalent to equal ability to control. Therefore, the fact that Mr. Altringer was unable to pilot the aircraft is immaterial.
The South Dakota cases do not provide more than a general abstraction of the joint enterprise doctrine.
Whether this analysis should apply to the flight of an airplane, where there was no question as to who was the pilot in command, and where the passenger was not trained to fly and could not operate the airplane, is somewhat questionable. Another uncertainty exists, as well, when, assuming a joint enterprise existed in the airplane flight, the negligence of Largent (failure to obtain full weather briefing) occurred before the putative joint enterprise trip commenced. Thus, we deem it questionable whether the pre-flight negligence could be imputed to Altringer, even assuming the joint enterprise existed once the flight commenced.
[1344]*1344However, notwithstanding these uncertainties, we find the legal analysis made by the trial court overlooks the overwhelming case law governing joint enterprise. There is no indication that South Dakota law is not in accord with the controlling legal analysis of other jurisdictions dealing with the law of imputed negligence and joint enterprise. In passing upon this issue, we are mindful that the courts do not favor the doctrine of joint enterprise and that parties alleging the existence of a joint enterprise for the purpose of imputing negligence have a heavy burden of proof. See Lester v. John R. Jurgensen Co., 400 F.2d 393, 396 (6th Cir.1968); Melville v. Maryland, 155 F.2d 440, 443 (4th Cir.1946); W. Prosser & P. Keeton, The Law of Torts § 72, at 522 (5th ed. 1985). As explained by Professor Prosser, the doctrine of vicarious liability arising from a joint enterprise rests upon an analogy with the law of partnership in which a mutual agency exists for purposes of conducting business, so that the acts of one are to be charged against another. Prosser, at 516. Similarly, a joint enterprise is akin to a partnership, but for a shorter period of time and a more limited purpose. Id. at 517. The law considers that each person associated in the enterprise “is the agent or servant of the others, and that the act of any one within the scope of the enterprise is to be charged vicariously against the rest.” Id.; see also Restatement (Second) of Torts § 491 (1979).
The district court found a joint enterprise between Largent and Altringer arising from their agreement to fly to California to inspect cattle purchased for their business. There can be no doubt that the purpose of this flight was directly related to their duties with Fall River Feedlots, Inc. It is their relationship to the corporation, however, that illustrates the inapplicability of the joint enterprise analysis to this case. Largent and Altringer’s relationship was not one of partners; they were employees and shareholders of a corporation. While the negligence of a corporation’s employees may be imputed to the corporation, it is equally clear that the negligence of an employee will not be imputed to a co-employee in the absence of tortious conduct of his own. See, e.g., Kocher v. Creston Transfer Co., 166 F.2d 680, 685 (3d Cir.1948) (existence of co-employee status does not, in and of itself, justify the appellation “joint enterprise”); Melville, 155 F.2d at 443 (relationship of fellow servants alone not sufficient for the imputation of negligence); Laird v. State Farm Ins. Co., 290 So.2d 343, 348 (La.App.), writ den., 293 So.2d 1984 (La.1974) (the fact that fellow employees are acting in the course and scope of their employment and in furtherance of their employer’s business does not in itself make them participants in a joint venture); Nadeau v. Melin, 260 Minn. 369, 382, 110 N.W.2d 29, 38 (1961) (even where passenger has right to control the manner of operation of a vehicle in which he is riding, the negligence of the driver, a fellow employee, will not be imputed to the passenger in the absence of a relationship which would establish vicarious liability); Connell v. Hayden, 83 A.D.2d 30, 443 N.Y.S.2d 383, 397 (1981) (a servant is not vicariously liable for the negligence of his co-employees; each is liable for his own negligence only); Bartholomew v. Oregonian Publishing Co., 188 Or. 407, 216 P.2d 257, 260 (1950) (mere co-employment is insufficient basis upon which to rest the imputation of a driver’s negligence to a passenger); Ratcliff v. Myers, 382 Pa. 196, 113 A.2d 558, 561 (1955) (it is well established that imputed negligence does not apply to co-employees engaged in a business mission for their employer); Restatement (Second) of Agency § 358(1) (1958) (agent not liable for conduct of other agents unless he is at fault for appointing, supervising, or cooperating with them); Restatement (Second) of Torts § 491, comment d (1979) (co-employees, by virtue of that relationship alone, do not engage in joint enterprise). Similarly, an officer of a corporation will not be responsible for the tortious conduct of another officer or agent of the corporation without some participation, authorization, or direction of his own in the tortious conduct.11
[1345]*1345The designated record before us reveals very little about the corporate structure of Fall River Feedlots. All we can glean from the record is that Largent was an incorporator of the company and, at the time of his death, was a shareholder and held the offices of president and manager. Altringer was a shareholder and assistant manager. As the district court found, this flight had the primary if not sole purpose of inspecting cattle purchased by Fall River Feedlots. In the absence of any other purpose for the flight that would meet the requirements of a joint enterprise under South Dakota law, we must find that Largent and Altringer were agents of Fall River Feedlots, not of each other. South Dakota law permits businesses to incorporate, and persons incorporating their business must be permitted to take advantage of that form of business. One advantage is limited liability of the corporation’s agents. In selecting the corporate form under which to operate the business, Largent and Altringer thus avoided the joint and several liability rules prevalent under South Dakota partnership laws. To now impose a joint enterprise relationship on Largent and Altringer would circumvent South Dakota protections pertaining to corporations. Under the above-cited principles, a finding that a joint enterprise exists would be impermissible.
Even if the corporate structure could be ignored, it is evident that as president and manager of the feedlot, Largent was Altringer’s superior. Thus, if any principal-agent or master-servant relationship existed, it would be with Largent as principal and Altringer as agent. Since, however, Largent was piloting the aircraft, it can in no sense be stated that Altringer had any right of control over the operation or direction of the plane. Thus, the element of mutual right of control necessary to establish a joint enterprise was not present in this case. See Robertson v. United Fuel & Supply Co., 218 Mich. 271, 187 N.W. 300, 301 (1922) (while the negligence of a servant driving will be imputable to a master, the negligence of a master driving is not imputable to a servant; the servant has no control over the master).
We vacate the judgment of the district court and hold that Altringer is not barred from recovery. We remand the case to the district court to determine damages and to enter judgment against the government for plaintiff Altringer. We remand to the trial court for reconsideration under the South Dakota comparative negligence statute as to whether Largent’s negligence was more than slight in comparison to the overall negligence of both of the government employees, Shields and Kasen.