In 1891, Corydon Purdy published a four-piece article in the Engineering News on “The Steel Skeleton Type of High Buildings.” Skeleton framing, as we understand it today, was about a year old; using some of the older and looser definitions floating around in the 1880s and 90s, it was maybe three or four years old. In other words Purdy, one of the best engineers for this particular type of work at that time, was describing the new frame technology for an educated audience of engineers, but one that might never have thought about the details of design for such a structure.
There were a number of responses, one of which I’ve excerpted above. Foster Milliken was an engineer, and the firm he ran with his brother (inherited from their father, who founded it) provided structural steel for, among others, the Singer Building. So he was in the same class of engineers as Purdy: he actually knew something about the topic.
He disagreed with a statement from Purdy that it was not clear whether Z-bar columns (H shapes built up from zees and plates) or Phoenix columns (built-up round and octagonal sections) were superior for combined compression and bending, as a column is subjected to in a moment-frame building. The sketches accompanying his letter are great, because they provide a clear path to splicing Phoenix columns and connecting beams to them, two things that are much more obvious with Z-bar columns. A “pintle” is a short, narrow pin that was used at the time to connect wood post and cast-iron columns, and he has recreated it in steel in the center of the phoenix section. For ordinary columns, the lintel consist of a pair of angles catty-corner to one another and connected by plates; for column with a lot of moment, the intel is a miniature phoenix column inside the phoenix columns. I particularly like the detail for providing skew connections when the supported beams are not parallel to the column-section axes.
New technologies always go through a period where a lot of new ideas are tried by people excited by the possibilities and, as many of those ideas are found to be flawed, the spread narrows down. Phoenix columns were a good idea for the era when all columns for heavy load had to be built up, but in the end they required more work similar (if slightly weaker) Z-bar, other built-up H, and built-up box columns, so they gradually were losing market share even before one-piece wide-flange columns were invented.
