564 So.2d 760 (1990)
David W. DODSON, et ux., Appellants,
v.
WEBSTER PARISH POLICE JURY, Appellee.
No. 21609-CA.
Court of Appeal of Louisiana, Second Circuit.
June 20, 1990.
Writ Denied October 19, 1990.
Nelson, Hammons & Johnson by John L. Hammons, Robert C. White, Shreveport, for appellants.
Lunn, Irion, Johnson, Salley & Carlisle by Charles W. Salley, Shreveport, for appellee.
Before FRED W. JONES, Jr., NORRIS and HIGHTOWER, JJ.
NORRIS, Judge.
In this wrongful death action the plaintiffs, Mr. and Mrs. David Dodson, seek damages for the death of their 15-year old son, Mark Dodson. Mark was killed in a one-car accident when his Jeep ran off a parish road and hit some trees. The plaintiffs sued the Webster Parish Police Jury, the custodian of the road, alleging the roadway was defective and caused the tragic accident. After a bench trial, the district court concluded the plaintiffs had failed to prove causation between the accident and the alleged defects. From the judgment dismissing their suit, the plaintiffs appeal. For the reasons expressed, we affirm.
Facts
The accident occurred on February 27, 1987 around 7:15 a.m. Mark Dodson was driving to school in his father's 1976 Jeep CJ5, which he had used regularly since he got his driver's license a few months earlier. He was taking the usual route from his home on Sugar Creek Drive; he drove south down Webster Parish Road No. 114 (called Germantown Road) to Minden, where Mark was in the 9th grade at Webster Junior High. The pavement was wet as there had been rain earlier. As Mark completed a gentle leftward curve about a mile north of Minden, he apparently lost control of the Jeep and ran off the left side of the road. Though there were no skid marks on the pavement (not unusual in wet conditions), tracks on the east shoulder showed that Mark slid 106' before hitting a tree, spinning counterclockwise, then hitting another tree. Mark was ejected from the Jeep and died almost instantly. The Jeep sustained heavy damage. There were no eyewitnesses.
Trooper C.D. Lee, who investigated the accident, testified that there were no defects on the road's surface, such as pot holes or ruts, that would have caused Mark to lose control. He also testified there was no physical evidence that Mark had strayed off the right edge of the pavement before careening to the left. Trooper Lee had never had any trouble rounding this curve and he had never investigated any other accidents there. He characterized the curve as "not very sharp," an observation that the numerous photos in evidence verify. On the basis of the damage to the Jeep, Trooper Lee estimated Mark must have been traveling 65 m.p.h. when he left the road, or 10 m.p.h. over the speed limit.
Mr. Jimmy Hart, Webster Parish's superintendent of roads, testified that his department attempted to look at each parish road at least once a month, though there is no formal policy for inspections. He testified that the parish did not keep records of accidents at any given location, but he knew of two or three accidents in that general area. Germantown Road, he said, was an old oxcart trail that was not blacktopped until about 1950. A second overlay was done in 1972 or 1973, when the width was increased to 24' because of heavier traffic; a final overlay was done in 1975. Though he had no statistics, he estimated about 200 cars used the road each day.
There was much evidence about a large hole and culvert opening on the right (west) shoulder and about improper banking of the road surface. The road's average width of 24' meets the applicable standards. In front of the curve is a "curve" sign without an advisory speed limit. Along the curve there is a grassy shoulder of uneven width. At the end of the curve, however, the right shoulder narrows to about two feet and drops suddenly into a large, 4-foot deep hole. At this point a culvert underlies the road; for some reason it extends 18' on the left side but only three or four feet on the right. It opens into the large hole. A striped "hazard" sign, measuring 12" by 36", stands seven feet north of the culvert and about two feet off the pavement. This culvert is about 75 to 100 feet south of where Mark's Jeep left the road.
Dr. William Hadley, a professor of civil engineering at Louisiana Tech, testified for the plaintiffs as an expert in traffic engineering, highway design, traffic safety and accident reconstruction. He surveyed the scene and prepared a series of helpful diagrams, admitted as Ex. P-13. Approaching this culvert from the north, Germantown Rd. forms a gentle leftward curve with a radius of 1,250'. The banking, which is important to offset centrifugal force in a turn, is called super elevation (expressed as an italic e) and is calculated by dividing the distance of vertical rise by the distance of horizontal run. He testified that the road had uneven e values; close to the culvert, the northbound lane had e = .048, an acceptable reading, but the southbound lane, in which Mark was traveling, had only e = .018, considerably less than applicable standards. With this low e, Dr. Hadley thought a driver traveling over 25 m.p.h. in Mark's lane would tend to pull to the right. Dr. Hadley also described the hole in the right shoulder as an "obvious" and "significant defect." He testified that this defect and a hazard sign so close to the pavement would cause a driver to "shy away" from the right edge of the road. Because of the inadequate banking, Dr. Hadley felt that a driver in the southbound lane, even if he was traveling at an otherwise safe speed like 45 m.p.h., would tend to stray to the right, perhaps off the road; but when he saw the hazard he would instinctively jerk to the left and lose control. Though he did not actually reconstruct the accident, Dr. Hadley felt that this interplay of factors should have been anticipated and remedied by the police jury. Mere signing was inadequate.
Dr. Olin Dart Jr., a consulting engineer from Baton Rouge and expert in the same fields, testified for the plaintiffs by deposition. He also surveyed the scene, and most of his measurements correspond with Dr. Hadley's. He stated that this portion of Germantown road should have e = .04-.08, and in this regard the road was substandard. Unlike Dr. Hadley, he conducted a "ball bank test" and concluded that 55 m.p.h. was a safe speed for this turn, at which a driver in Mark's lane should not have lost control. Dr. Dart was not sure that the banking contributed to the accident. He was emphatic, however, that the narrow shoulder, improperly placed culvert, deep hole and hazard sign created a major hazard on the road. Dr. Dart assumed that Mark had actually strayed off the right side of the road, spotted the hazard sign or the culvert, and then steered back onto the pavement too swiftly, causing the skid. He did not calculate Mark's speed; he assumed he was going about 45 m.p.h. when he lost control. He commented that Jeeps were "notorious" for poor handling.
Dr. Ned Walton, a licensed engineer from Bryan, Texas, testified for the police jury as an expert in highway design, traffic engineering, accident reconstruction and human factors. He conceded that the culvert and hole were close to the road, but he thought the signs provided adequate warning. He also conceded there was a minor banking deficiency, but said this should not have been a problem except under icy conditions. Like Dr. Dart, he performed a ball bank test; he also used a standard formula to calculate a safe speed. In this formula he used a coefficient of friction for a wet surface, which creates more friction than an icy surface; Dr. Hadley's calculations had been based on almost frictionless icy conditions that were not present on the date of the accident. Using the coefficient of friction for wet pavement and e = .025 (slightly higher than Dr. Hadley's), he found that 57 m.p.h. was a safe speed for this curve. He disagreed with Dr. Hadley's estimate, stating that he drove through the curve at 55 m.p.h. with no difficulty. He also disagreed with Dr. Hadley's theory of the accident; the fact that Mark left the road some 75' before the hazard sign belies any "near miss" of the hazard; at that distance, the culvert and hole are not visible. He thought the road was not unreasonably dangerous and that nothing on the roadway, either the banking, culvert, hole or shoulder, caused the accident. Instead, he attributed the accident to improper control of the Jeep, both by excess speed and poor lane usage.
All experts testified as to the applicable standards for highways, which the Department of Transportation and Development is charged to promulgate. See LSA-R.S. 48:35. The American Association of State Highway and Transportation Officials ("AASHTO") regulations, adopted in 1954 and revised in 1965, apply to all state and parish roads newly built or largely reconstructed after those effective dates. Dr. Hadley testified that Germantown Rd. did not meet the standards. Dr. Dart would not positively state that AASHTO standards governed this roadway, which antedated the standards. Dr. Walton testified that all work on Germantown Rd. since 1950 had been minor and the police jury was not bound by AASHTO.
Before trial, the district judge apparently told the attorneys that he would like to view the scene. During a break at trial on June 14, 1989, the plaintiffs' attorney suggested that the judge and attorneys visit the scene then. R.p. 169. They did so. Numerous photographs were also introduced.
Action of district court
The district court prepared written reasons for judgment. The court noted the expert testimony was "at some variance" in proving that Germantown Rd. posed an unreasonable danger. Without recapping the conflicting expert evidence, the court stated that its personal inspection of the accident scene was persuasive to resolve the issue; the road did not "create a trap for the unwary."
The court also found no evidence that the condition of the road, defective or not, caused this accident. The plaintiffs' theory of how the accident occurred struck the court as speculative and not supported by a preponderance of evidence. Because there was no causation, the court rejected the plaintiffs' demands.
Discussion
By two assignments of error, appellants urge the trial court was plainly wrong in failing to find that Germantown Rd. was unreasonably hazardous at the point of the collision and in failing to find that the condition of the road caused or contributed to the accident.
In a case arising from an allegedly defective highway, the basis of liability is LSA-C.C. arts. 2315 and 2317, which provide in part:
Art. 2315. Liability for acts causing damages
Every act whatever of man that causes damage to another obliges him by whose fault it happened to repair it. * * *
Art. 2317. Acts of others and of things in custody
We are responsible, not only for the damage occasioned by our own act, but for that which is caused by the act of persons for whom we are answerable, or of the things which we have in our custody. * * *
Under both articles, liability hinges on whether the defendant has breached a duty to the plaintiff. Although the bases of liability are conceptually different, as negligence need not be proven in a strict liability case, the actual duty is the same. Manasco v. Poplus, 530 So.2d 548 (La.1988); Kent v. Gulf States Util. Co., 418 So.2d 493 (La.1982). As the custodian of Germantown Rd., the police jury owes a duty to motorists to keep the road and shoulders in a reasonably safe condition. LeBlanc v. State, 419 So.2d 853 (La.1982); Sinitiere v. Lavergne, 391 So.2d 821 (La.1980). Whether the custodian breached this duty depends on the unique facts and circumstances of the case. Myers v. State Farm Mut. Auto. Ins. Co., 493 So.2d 1170 (La. 1986); Manasco v. Poplus, supra.
Proof of a duty with its attendant risk of harm, and breach of that duty, is not enough to create liability in the defendant. The plaintiff must also prove by a preponderance that the defendant's conduct was a cause in fact of the harm suffered. Manasco v. Poplus, supra; Jones v. City of Baton Rouge, 388 So.2d 737 (La. 1980); Dixie Drive It Yourself v. American Beverage Co., 242 La. 471, 137 So.2d 298 (1962).
In finding or refusing to find defect and causation, the trial court has great discretion; the findings will not be disturbed absent a showing of manifest error. Arceneaux v. Domingue, 365 So.2d 1330 (La.1978); Towns v. Georgia Cas. & Sur. Co., 459 So.2d 124 (La.App. 2d Cir.1984), and citations therein. Appellate deference to the trier of fact's findings is based on the trier's superior position to assess the credibility of witnesses and the weight of expert documents. Canter v. Koehring Co., 283 So.2d 716 (La.1973); Virgil v. American Guar. & Liability Ins., 507 So.2d 825 (La.1987). Reasonable assessments of credibility should almost never be disturbed on appeal. Rosell v. ESCO, 549 So.2d 840 (La.1989).
On the question of defect, there was ample evidence to show that if Germantown Rd. were a recent construction it would fall short of AASHTO and DOTD standards in various particulars. The shoulders are too narrow, the culvert too near the pavement, and the super elevation irregular. The evidence clearly shows, however, that the road antedates the current standards. Thus evidence of noncompliance or failure to repair does not prove the existence of a hazardous defect. Myers v. State Farm, supra; Manasco v. Poplus, supra. The question is simply whether the road posed an unreasonable risk of harm where the accident occurred. Dill v. State DOTD, 545 So.2d 994 (La. 1989).
The plaintiffs' expert testimony surely asserted a defect, but when closely scrutinized and stripped of unsupported assumptions it is not as strong as at first blush. Dr. Hadley cited inadequate super elevation but did not make a reasonable allowance for friction; the particular effect of poor super elevation did not occur. He said the culvert and hole posed an apparent threat, but Dr. Walton flatly disagreed and Dr. Dart was equivocal at best. He also said the striped hazard sign, placed too close to the road, would have the same effect of making Mark veer to the left, but this effect must have been minimal on someone who had driven the road so often. Dr. Dart's findings and theory were similar, but he freely admitted his reliance on an assumption that Mark had first strayed onto the right shoulder, an assumption that was not proved. Under close questioning, he also admitted that the super elevation was not necessarily a problem, if Mark was traveling at as moderate a speed as he assumed.
By contrast, Dr. Walton admitted the minor problems but thought the road was not unreasonably dangerous. Neither he, Trooper Lee nor the plaintiffs themselves claimed they had ever had any difficulty rounding the curve. There was evidence of some accidents in the general area but not at this spot. The judge's personal inspection, accompanied by both counsel, corroborated that this stretch of road was not unreasonably dangerous.
In brief appellants contest the judge's trip to the scene as prejudicial error. There was no objection to the judge's action so the error is considered waived. LSA-C.Ev. art. 103A(1). In fact, plaintiffs' counsel acquiesced and suggested the time to go. Moreover, the judge's personal observation did not supersede the trial evidence or usurp the litigants' role in presenting the case. It was merely an effort, within the trial court's discretion, to clarify evidence that was in conflict. Hardy v. Brown, 372 So.2d 573 (La.App. 2d Cir.1979); Landry v. Jefferson Davis Parish Sch. Bd., 478 So.2d 194 (La.App. 3d Cir.1985). Under the circumstances it was not error.
The case of Dill v. State DOTD, supra, cited by appellants, is distinguished. There the plaintiffs adduced ample proof that the River Road posed severe hazards of long standing, including narrow lane width, severe curvature and a deteriorated lane of road surface. That evidence was sufficient to convince the trial court of DOTD's fault (but not enough for the Supreme Court to affirm the 100% allocation). In the instant case the evidence as to defect is not nearly as persuasive. The district court's finding is not manifestly erroneous and this assignment lacks merit.
Apart from the question of defect, the court found no proof of cause in fact. Was this plainly wrong?
Common sense dictates that a 4' deep hole in a road shoulder is a hazard and would cause damage to any motorist unfortunate enough to tumble into it. This is not what happened; Mark left the road considerably ahead of the hole, about 75', which is too far to be a "near miss." R.p. 235. The theory that he took sudden, evasive action in response to the culvert, hole and hazard sign is not supported by the evidence. The anticipated effect of poor super elevation of the right lane did not occur either; Trooper Lee testified that if Mark had first run onto the right shoulder, he would have noticed and recorded that fact. R.p. 168. To urge that this happened is, as the trial judge noted, speculation. The tragic but no less reasonable hypothesis is that Mark was inattentive and took this curve partly in the inside lane and at an excessive speed. These acts were independent of the condition of the road. In accord with expert testimony, his speed and placement caused him to veer out of control to the left. The plaintiffs did not convince the district court that the condition of the road, more likely than not, caused the accident. The court was not plainly wrong in crediting the defense's theory over the plaintiffs' and in refusing to find causation.
For the reasons expressed, the judgment is affirmed at appellants' cost.
AFFIRMED.