"Washington will get a fitting portal," intones one document describing a proposal to develop a 10.4-acre site in the District near the 14th Street bridges. Another proposal for the same piece of land claims it will "draw the neighborhood back to its original connection to the Mall and the waterfront." Its "signature project," says yet another development firm, "will become known throughout the city, the region and the nation."

These and many other blasts of hot air were emitted last week from a crowded Washington conference room where members of the board of the District's Redevelopment Land Agency met to interview five teams of developers competing for the right to build upon the so-called Portal Site, the last remaining slice of the Southwest urban renewal pie.

The cause of this crescendo of inflated rhetoric was not hard to scent. Megabucks were in the air as thick as rush-hour fumes. But what the developers actually were talking about is an enclave of office buildings and, regrettably, the designs they actually proposed rarely if ever scaled the heights promised by their words or matched the amounts of money -- in the $300 million range -- they seem content to spend.

Part of the problem clearly is the location and nature of the site -- nothing more than a couple of huge parking lots on gently sloping land bounded on the west by 14th Street, on the east by 12th Street, and north and south, respectively, by the formidable wall of Agriculture Department buildings along D Street and a challenging barrier of freeways and roads bordering the Washington channel.

To call this site a portal, an entranceway or a "major gateway" is to stretch the meaning of the words. The site is hard to get to by car and unpleasant to reach on foot. There is, to be sure, a subway stop on Independence Avenue a block or so away, but the site is almost sealed off from L'Enfant Plaza to the west and from the waterfront and the residential communities of Southwest. As if this were not enough, it is split right down the middle by a parallel pair of functioning railroad tracks.

This imposing set of obstacles and other constraints produced a certain sameness in the five proposals. Each foresees construction of a series of office buildings with retail components on both the northern and southern segments of the site and a hotel in its southwest corner (the one with the most desirable view looking toward the Jefferson Memorial).

There has been some scattered talk about somehow including housing in the mix of proposed uses, but this is a late, uneconomic and foolhardy idea. Housing, at least in the density required to make it work, doesn't really fit the site, which in fact is one of the few places left in Washington where putting up a cluster of new office buildings makes great sense.

In terms of urban design, the first and chief decision each architect had to make was what to do with the Conrail tracks that cut through the site on the diagonal that L'Enfant envisioned as Maryland Avenue -- radiating from the Capitol like a great southern echo of Pennsylvania Avenue. This decision was crucial not only to the view of the Capitol offered by this site, but also to the way the new buildings relate to each other.

Two distinct views emerged on this issue. One was to cover the tracks as much as possible. The other was to leave them as they are and tidy them up with grass and trees. Two firms favor the hands-off approach: Welton Becket Associates, working for the Rockefeller development team, and Bryant and Bryant for the Arthur Winn-Dominic Antonelli group. (Credits here are terse, otherwise they would run book-length.) The cover-and-build idea attracted Vlastimil Koubek for the group headed by Theodore Lerner and Melvin and Edward Lenkin; Skidmore, Owings and Merrill for the Conrad Cafritz team, and Arthur Cotton Moore for Western Development.

The cover-and-build approach wins this contest hands down. It just doesn't make sense to pay so much attention to the Maryland Avenue vista when the avenue in fact is a railroad track . . . and what motorist coming into the city would dare slow down to admire the view? Furthermore, connections between the two halves of the project become tenuous at best -- the Bryant and Bryant scheme proposes a solitary, off-center bridge, while Welton Becket's provides an unsatisfactory tunnel.

Covering the railroad tracks, on the other hand, preserves the Capitol view for those who can pause to enjoy it -- hotel guests and users of the office enclave -- and allows for more enlightened organization of outdoor spaces on the site. By building a platform over the tracks,the architects can at least try to transform a necessary nuisance into a landscaped urban amenity and to establish the spirit of a city boulevard on the site.

Each of the schemes has significant drawbacks, however. Koubek, assisted by Sasaki Associates in the landscape design, proposes a nice balance of paved and green spaces. But the east-west alignment of his ponds, fountains and terraced buildings goes contrary to the diagonal axis of the vista. Moore's idea of a central, circular space reinforced by curved building facades has a notable touch of that old Washington grandeur, but it is a wide, hard space with too little grass and too few trees, and his acknowledged theft of Robert Venturi's city map idea for the Western Plaza on Pennsylvania Avenue is too de'ja vu.

In terms of open space planning, Skidmore, Owings and Merrill's plan strikes a good balance, combining a circular viewing platform at the western corner and an elegant and accessible street on the southern portion of the site. But then, the firm's notions of architectural style -- clean-lined contemporary, relieved only by crisp echoes of Beaux Arts detailing -- provides nothing more than a slick, elegant, very corporate surface for this big cluster of buildings. Handsome it might be, but it is also tired, repetitive, dull, off-putting and quintessentially safe.

In fact, with the exception of Moore's almost off-the-wall historicism -- the architect refers to it accurately as a "modern/classic/Beaux Arts" approach to design (in other words, a pastiche) -- the architecture throughout this competition lacks spirit and imagination. Koubek's scheme for windows and surfaces is a lesson in unrelieved tedium and his circular hotel belongs in some ravaged resort. Welton Becket's tastefully mansarded cluster of buildings is but a weak echo of Washington's traditional federal office buildings (and the weaker, 1930s versions at that). Bryant and Bryant's scheme for a Portmanesque glass-covered court and a hotel with honeycomb balconies is an inappropriate cliche'.

This leaves Moore standing pretty much alone, flaws and all. His scheme has a lot of circus hokum in it: rounded pavilions at the building corners, with curiously proportioned domes atop slim paired columns; a huge decorative arch at the waterfront side framing a tiny carousel; a curiously isolated, pedimented entranceway on the northern facade; towering theatrical windows in places, grandiose entrance marquees and lots more. A cacophonous pastiche, a folly -- Moore's scheme, nonetheless, possesses an underlying functional intelligence, and it promises something that the other proposals surely cannot deliver, and that is a level of visual excitement that might, just might, become memorable.

The last thing we need in Washington is another four-block-big exercise in architectural caution.