Clayton H. Shrout, Omaha, Neb., on briefs, for appellant.

John B. Henley, Colorado Springs, Colo. on brief, for appellees.

Before BRIGHT and HENLEY, Circuit Judges, and HARPER, Senior District Judge.*

HARPER, Senior District Judge.

Appellant, School District # 54, brought this diversity action to recover damages incurred as a result of the failure of the roof of its newly constructed high school. Suit was originally filed against the school district's architect, the general contractor, the roofing subcontractor, the roofing material manufacturer, Celotex Corporation (hereinafter referred to as Celotex), and the roofing bond surety, Aetna Casualty Company (hereinafter referred to as Aetna). However, settlement agreements entered into prior to trial left Celotex and Aetna as the sole parties defendant in the cause, which proceeded to trial before the Court.1 The lengthy trial consisted entirely of appellants' case which was based upon theories of misrepresentation, negligence and breach of warranty, both express and implied. At the close of plaintiffs' case the defendants rested without introducing direct evidence. The district court entered judgment for defendants, which judgment we now affirm on appeal.

In order to best understand the nature of the present litigation it is advisable to first consider constructional aspects of a large roofing system, only then proceeding to the facts specifically applicable to the controversy. The primary structural component of a roof system is the structural deck, also known in the industry as the substrate. This deck, which may be composed of a variety of materials, is the structural member which bears the load of the insulation and waterproofing.

It is the waterproofing alone which protects both the interior of the building and the substrate itself from water which alights upon the roof. The substance applied to waterproof the substrate is referred to by its generic name of bitumen, and it may be either a tar or an asphalt. This bitumen, be it tar or asphalt, lacks the elasticity required to accommodate normal movement of the structural deck, either thermal or moisture induced. Accordingly, in order to enable the bitumen to resist structural movement and maintain its efficacy as a waterproofing agent, layers or plies of felt are attached to the substrate. These felts, which are not in themselves watertight, are referred to collectively as a roofing membrane and serve as a substrate to an application of tar or asphalt. This membrane distributes the stress imposed by the movement of the structural deck and resists the contraction forces which the bitumen itself induces.

Prior to 1964, the use of a four-ply built-up roofing membrane was common in the construction industry to achieve a watertight roof. This system was and is comprised of a base sheet to which are attached three plies of felt impregnated and bonded together with asphalt. In 1964, a two-ply roofing membrane system, commonly known by its tradename "Bond Ply", was developed by the predecessor in interest of Celotex and introduced in the roofing industry. This new system, which consisted of but two plies of felt impregnated and bonded together with asphalt, was represented as the equivalent of the four-ply membrane in performance with the advantages of lower cost and greater ease in installation.

In 1967, appellant, a school district located in the suburbs of Omaha, Nebraska, engaged an architectural engineering firm to design a new high school. The school was designed and built as a large structure with several distinct roof areas. The roof over the classrooms and administrative offices is of a complex domed geometry which incorporates multiple valleys, ridges, gabled areas and compound curves. The school's gymnasium is covered by a roof considered more conventional in architectural design. In all areas of the building the roof is supported by structural steel upon which fiberglass sheeting and formboard rest and over which a gypsum substrate has been poured.

The architect employed by appellant, found by the trial court to be relying at least in part on material published by Celotex, specified the "Bond Ply" system as the roofing membrane to be used on the high school. Prior to the installation of the "Bond Ply" system the project's roofing subcontractor applied to Celotex for approval of the roof specifications to insure that they were sufficient to entitle the roof to a bond. Celotex modified these specifications to provide for attachment of the roofing membrane to the underlying structural deck by a process of liquid adhesion known as "mopping", rather than by nailing as had originally been specified. Though the manager of the roofing subcontractor expressed his dissatisfaction with this modification, on the insistence of Celotex the first ply of the system was solidly mopped to the gypsum deck.

Subsequent to the completion of the high school it was discovered that massive leaks had developed in its roof. These leaks, which were the result of splits in the roofing membrane, caused extensive damage to the interior of the building. After it became apparent that partial or limited repairs were inadequate to correct the leakage problem, it was determined that the Celotex "Bond Ply" membrane system would be replaced with a conventional four-ply system. Appellant then sued for the cost of replacement, interior property damage and interruptions within the building during its use.

In holding for the defendants the district court concluded that evidence indicating that the Celotex "Bond Ply" roofing system was inherently defective or that any defect in the membrane caused the leaks was skeletal at best and did not preponderate in favor of appellant. With regard to causation the Court specifically found that, as constructed, the school building's roof was subject to excessive expansion and contraction of the gypsum substrate and the underlying structural steel. The two-ply membrane was not found to have split due to inadequate strength but rather because its elasticity was not great enough to accommodate the strain imposed upon it by the structural deck. Here the district court noted that, although the evidence suggested the two-ply system was not as strong as the four-ply system, nothing indicated that it was substantially less elastic. Considering the contention of the school district that Celotex negligently modified the architect's specifications for the attachment of the membrane, the district court held that the solid mopping of the system to the gypsum deck was not unreasonable in light of industry practice prevailing at the time of the school's construction.

By its briefs to this Court on appeal appellant has argued that in review of the decision of the trial court we are not bound by the "clearly erroneous" standard set forth in Rule 52(a) FRCP. It is appellant's contention that Rule 52(a) is not applicable in a review of those findings of a trial court which are drawn from nondemeanor evidence since the credibility of a witness is not in issue. As applied to the instant case, appellant maintains the clearly erroneous standard is not appropriate since appellees neither presented direct evidence at trial nor contradicted appellant's witnesses, and because the trial court, in finding Celotex did not negligently modify the roofing specifications, relied exclusively on documentary evidence.

We do not agree that the clearly erroneous standard of Rule 52(a) is limited in application in the manner suggested by appellant. Rule 52(a) provides in relevant part:

Rule 52. Findings by the Court.

(a) Effect. In all actions tried upon the facts without a jury or with an advisory jury, the court shall find the facts specially and state separately its conclusions of law thereon, and judgment shall be entered pursuant to Rule 58; * * *. Findings of fact shall not be set aside unless clearly erroneous and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses. * * *.

Though Rule 52(a) does not by its terms specifically provide for review of findings of fact based upon nondemeanor or uncontested evidence, this Court has held that the clearly erroneous standard, "applies to reasonable inferences to be drawn from stipulated or undisputed facts, and it is for the trial court rather than the appellate court to draw legitimate and permissible inferences." Lewis v. Super Valu Stores, Inc., 364 F.2d 555, 556 (8th Cir. 1966). See also United Stores of America, Inc. v. Insurance Consultants, Inc., 468 F.2d 1010, 1011-12 (8th Cir. 1972); St. Louis Typographical Union No. 8 v. Herald Co., 402 F.2d 553, 557 (8th Cir. 1968).

This case, which among other things requires evaluation of expert witnesses' testimony, is unlike those involving construction of an unambiguous written agreement, e. g., Frito-Lay, Inc. v. So Good Potato Chip Co., 540 F.2d 927, 929-30 (8th Cir. 1976); Ralston Purina Co. v. Hartford Acc. & Indem. Co., 540 F.2d 915, 918 n. 3 (8th Cir. 1976), in which we have held that the clearly erroneous standard of Rule 52(a) does not apply.

Accordingly, the standard of review to be utilized in this appeal is that enunciated in Zenith Radio Corp. v. Hazeltine Research, Inc., 395 U.S. 100, 89 S. Ct. 1562, 23 L. Ed. 2d 129 (1969):

In applying the clearly erroneous standard to the findings of a district court sitting without a jury, appellate courts must constantly have in mind that their function is not to decide factual issues de novo. The authority of an appellate court, when reviewing the findings of a judge as well as those of a jury, is circumscribed by the deference it must give to decisions of the trier of the fact, who is usually in a superior position to appraise and weigh the evidence. The question for the appellate court under Rule 52(a) is not whether it would have made the findings the trial court did, but whether "on the entire evidence (it) is left with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 542, 92 L. Ed. 746 (1948).

395 U.S. at 123, 89 S. Ct. at 1576. See also Brown v. United States, 419 F.2d 337, 340 (8th Cir. 1969).

With these principles in mind we have examined the lengthy record presented by this litigation and find ourselves unable to conclude that the district court erred in finding that: (1) the proximate cause of the membrane splits resulting in the leakage was excessive structural movement in the building rather than an inherent defect in the "Bond Ply" product; and (2) Celotex did not act negligently in modifying the specifications for the attachment of the roofing system to the gypsum structural deck.

Though no attempt was made, by the use of meters or otherwise, to measure structural movement within the school, the designated record presents abundant support for the finding of the trial court that the building experienced abnormal and excessive structural movement. This fact, considered with the absence of any evidence indicating that the "Bond Ply" system was less elastic than the traditional four-ply system, substantiates the holding that structural movement rather than an inherent defect in the membrane system was the proximate cause of the damage sustained.

The record reveals both direct and circumstantial evidence pointing to structural movement as the cause of the splits which appeared in the roofing membrane. It is first to be noted that the very design of the school building created the potential for abnormal structural movement and stress within the building. The manager of the roofing subcontractor, Chris Wille, testified that the structural steel which suggested the walls of circular and irregularly shaped rooms was placed in many directions and necessarily resulted in movement within the building in many different directions. Milton Olson, a roofing expert employed to correct the leaks in the roof, deposed that he was thwarted by structural movement in his attempts to repair substrate cracks which had appeared directly beneath splits in the roofing membrane. These movements invariably caused recracking. Olson further explained that each dome was constructed in such a manner that its base could not expand, with the result that thermal movement in the dome was channeled in an upward direction. He stated that cracks present in the domes were evidence of such upward movement and that no roofing membrane could sustain those kinds of stresses.

Dennis McNeil, a branch manager of a roof and waterproofing construction firm engaged by appellant to repair the roof, ratified Olson's analysis by his admission on cross-examination that cracks around the perimeters of the domes and at right angles to those perimeters could be characterized as classic structural movement cracks. McNeil also acknowledged his statement made on deposition to the effect that splits in the roofing membrane on ridges located between the domes were an indication of some movement occurring at those points.

Various efforts to minimize the effects of structural movement within the building which were taken during the replacement of the "Bond Ply" roofing membrane are additional evidence of the existence of such movement in abnormal amounts. New expansion joints were installed to protect the replacement membrane from the splitting caused by the movement of the building. McNeil testified that these new expansion joints were placed between the domes for the purpose of accommodating structural movement in those areas. It was his opinion that expansion joints should probably have been so located in the original construction of the school. Olson deposed that, due to cracking along the top edges of the ridges on sloped roofs, new expansion joints were installed there with the effect that all such ridges were made into a "complete expansion joint." Nelson Hymans, a structural engineer with a consulting firm retained by appellant, testified that in the installation of the replacement four-ply system, unlike the original, a number of roofing membrane contraction joints were located to accommodate any shortening of that membrane and thereby protect it from splitting. It was his opinion that if the stress of contraction were not accommodated, any membrane, be it two-ply or four-ply, would either stretch or split.

Efforts were also taken to reduce the amount of structural movement through the installation of insulation between the gypsum substrate and the replacement membrane in those areas where the "Bond Ply" system was removed. Olson deposed that the insulation would relieve stresses emanating from the gypsum deck onto the roofing membrane and would reduce the movement of the structural deck under thermal changes. It was his opinion that such isolation of the membrane from the main deck system was absolutely necessary to diminish the amount of thermal expansion in that deck which would ultimately be transferred to the roofing membrane. This opinion was affirmed by the testimony of McNeil that the original roofing specifications should have included the installation of insulation for two reasons. First, isolation of the membrane system would protect it from shrinkage cracks developing in the gypsum deck. Second, the insulation would reduce the effect of variations in temperature on the gypsum deck.

Pertinent to the issue of causation, though not mentioned by the trial court in its memorandum opinion and not necessarily indicative of the existence of excessive structural movement in the school, is evidence that the roof of the building was improperly constructed. Unworkmanlike installation of the roof in two particulars may well account for, and at the least contributed to, the development of the leaks which eventually required that the "Bond Ply" membrane be replaced. Testimony was adduced which indicated that ponding, which results from improper sloping in the roof, occurred in the relatively flat area between the gymnasium and the domes and in the area of the domes themselves. Leaking took place in both of these locations. Upon opening up the roof for inspection, Olson observed felt plies which had deteriorated to the point where a substantial amount of water was entering the school. This deterioration was attributed to the lack of proper drainage. McNeil testified that asphalt roofing membranes, like both the "Bond Ply" system and its four-ply replacement, will gradually deteriorate under conditions of ponded water. These ponding problems required that the roof be built up to attain a slope which would provide proper drainage. The second instance of improper roof installation is revealed by the testimony of Hymans and Olson, who both observed several original expansion joints with holes and gaps which allowed water to enter the school building.

Having found that excessive structural movement within the building was the proximate cause of the membrane splits which resulted in the leakage of water, the district court proceeded to consider the liability of Celotex for its purported negligence in requiring that the "Bond Ply" system be solidly mopped instead of nailed to the substrate. The district court found that, in light of prevailing industry practices at the time of construction, it was not unreasonable to mop a roofing membrane to a poured gypsum deck. Though the appellees contend the question of negligence has been mooted by the holding with respect to the proximate cause of the damage, we shall briefly address the issue.

At trial appellant produced Edward Schreiber, the president of a roof and waterproofing consulting firm, as an expert witness in its behalf. Schreiber, when asked his opinion on the effect of the solid mopping of a built-up roof, responded in the following manner:

Well, what the effect is, by being intimately bonded to what I term an acrobatic substrate, that the stresses that are developed in the normal working and splitting of that substrate will exceed the tensile strength that is available in the membrane to resist, and you will get a splitting. Now that will occur whether there is movement in the substrate or whether there is a crack in it initially. But those stresses added to the normal stresses that occur in a membrane will result in the fracture of that membrane.

He further opined that nailing the first sheet of the membrane to the structural deck, as one of the "classically accepted" methods of attachment, would provide for the divorce of the two components and thereby avoid the concentrated impact of structural stress experienced with solid mopping.

It is to be noted, however, that the actions of Celotex are not to be judged in the light of practices presently prevailing in the roofing industry. The testimony of Schreiber, based as it is on present knowledge and practice, is not necessarily persuasive with regard to the reasonableness of actions taken eight years earlier. The trial court recognized this when it chose to rely upon the testimony of Olson to establish that, at least at the time of the school's construction in 1968, the solid mopping of a roofing membrane was not universally considered inadvisable. When asked if in his opinion it made any difference that the first ply of the "Bond Ply" roofing membrane had been mopped instead of nailed, Olson answered:

A There's a big argument in the roofing industry as to which is the better of the two, nailing or hot mopping of the membrane. They've both been satisfactory in many instances, and the only problem that a lot of contractors over the country have noticed in nailing is that the nails eventually work up and come through, through the membrane, and therefore, most of them would much rather hot mop than

Q All right.

A put it down with nails, although, all of the specifications now, and since about the time of this application, they come out with nailing as the way to do it.

Q They've changed form mopping to nailing?

A Yes. They used to mop or nail, you had your preference.

Olson's deposition also reveals that 1,100,000 square feet of solidly mopped roofing was installed and has for some time been satisfactorily serving as a watertight sheathing for a Western Electric plant located in Omaha. Though the Western Electric roof was described as a five-ply system, we are satisfied that this evidence, along with Olson's testimony regarding the evolution of attachment practices in the industry, establishes that the district court was not mistaken in finding that Celotex did not act negligently in insisting that its "Bond Ply" membrane be solidly mopped to the substrate.

The judgment of the district court is affirmed.