If you visit the "lathe Archive" and go to the "Hendey" subsite, there are photos of the Hendey plant. These include some shots taken in the pattern shop & foundry.
The lathe bed pattern for a typical lathe bed would have been a "loose piece" pattern, having three (3) major sections:
-base flange (if used)
-bed body with core-prints (more about that in a few lines)
-ways
A stack of molding flasks would be used so that there were three sections (drag, middle "cheek", and cope on top).
The main body of the bed would be molded in the cheek or middle flask section, with parting lines where the loose pieces of the pattern for the bedways and base flange joined the main body of the bed.
Core prints were used to establish locations of the dry sand cores for the "girths" of cross members and openings that spanned inside the two vertical sides of the bed casting.
Dry sand cores were used to mold the girths (which were often like an inverted "U" channel section) or to mold the "lattice webbing" (used on South Bend's heavy 10" lathe beds).
Dry sand cores might also be used on the side of the bed to create a relief for the feed shaft and lead screw, and to include the raised lettering and any mounting pads for the quick change gearbox and rear bearing block for the feed shaft & lead screw,
Dry sand cores enabled intricate castings with otherwise "unmoldable" geometry to be made. Core prints are usually some sort of boss or projection that sticks out from the body of a pattern. The core print is a means of locating and fixing a dry sand core in the mold cavity.
Molding a lathe bed was a fairly tricky business as, once the main pattern parts were withdrawn from the sand, the cores had to be set. The core locations had to be checked in the mold, as cores, despite a core print, could cock. Properly venting a mold with dry sand cores was also a bit trickier than a regular sand mold. Then, there was the matter of providing proper gates and runners and risers to prevent shrinkage problems. A lathe bed is a long casting with deep, relatively thin side-walls and heavy sections at the top and bottom, as well as the girths or webbing on the inside. Arranging the gates and runners so the molten iron filled the mold uniformly without one section getting ahead of the other, and arranging for risers to prevent shrinkage defects also took some doing.
The other part of the equation is the molding sand and the "facing". Regular green molding sand uses silica sand and clay. Silica sand melts and vitrifies (forms glass) at some temperature below the temperature of molten iron. If molten iron is poured into a sand mold, the heat of the molten iron will vitrify the sand closest the iron and create a glassy scale and all sorts of defects in the casting (inclusions of this glass and porosities), along with a very hard shell on the casting. This problem was recognized at least a couple of hundred years ago. The problem was solved with what is known as a "facing" on the sand mold. This is often nothing more than powdered soft coal, known in foundry work as "plumbago". This is dusted onto the mold and cores before the mold is closed for pouring. When the mold is poured, the molten iron comes into contact with the plumbago. Being made from soft (bituminous) coal, the plumbago dust promptly forms coal gas. This coal gas forms an insulating space between the molten iron and the molding sand. The coal gas dissipates thru the venting in the sand mold. But, this coal gas "barrier" remains just long enough for the iron to "skin" or "freeze" on its surface. This lowers the temperature below the melting point of the sand.
Pouring iron and semi steel castings is a trickier business than molding lower melting point metals such as aluminum or bronze.
The other requirements for a pattern are to have sufficient allowance for shrinkage of the iron. Patternmakers used what was known as a "shrink rule" which had the shrinkage allowance built into the graduations. Different shrinkage rates for different metals. Iron, as I learned as a kid in Brooklyn Tech HS in the 60's, shrinks 1/8" per foot.
When a pattern is made, aside from the allowance for shrinkage, there has to be some consideration or sense for how the molten iron will cool and what stresses will be developed. Every corner on the pattern has to be rounded on external corners, and every "internal" corner has to have a generous fillet.
Then, we have to think about draft, or the relief angles on the pattern so it can be withdrawn from the sand. There are general rules for draft on patterns, but for something with cores and complex geometry, extra draft is usually put on the patterns.
Finish allowance is yet another issue. Normal finish allowance for machining is 1/8". Finish allowance has to be included on any surface that is going to be finished by machining and may be a bit more on some surfaces where there may be intersecting surfaces and perhaps some deeper cuts meeting shallower cuts.
Then, we get to the age-old debate. Patternmakers and molders, at least years ago, insulted each other and each craft claimed the other did not know what the ---- it took to make a good casting. The molders accused the patternmakers of making patterns which were next to impossible to mold or get a good casting with, and the patternmakers accused the molders of being prima donnas and being unable to mold anything remotely complex. Add to this the foundry's requirements. The foundry will have certain sizes and types of molding flasks, certain types of sand they use, and their own ideas and preferences for what a pattern needs to have.
Finally, we get to the alloy of iron or semi-steel the foundry will pour. Years ago, lathe beds were poured with iron melted in a cupola furnace, using a coke fire. The iron poured was often nothing more than busted up scrap- worn our machinery castings, busted up engine blocks, old cast iron pipe fittings and radiators. This was a basic gray cast iron. Sometimes, fresh blast furnace pig iron ingots were added to the melt. This produced a fairly fine grained casting, but it did not have much tensile strength and when machined, was fairly soft. This meant a lathe bed would wear more quickly. The machine tool builders started ordering "semi steel castings". These were castings having a certain percentage of scrap steel in the melt. This added strength and increased wear resistance. It let them cast lighter lathe beds. Then, the higher strength nodular and ductile irons came into being. These are much more predictable, more machineable and have much higher strength and a better grain structure. It's all a question of what the foundry you contract with to pour the lathe bed is melting for their castings. Some cast irons are a bitch to machine. Some are more dimensionally stable than others.
After casting comes snagging (grinding off the extra iron in the form of gates, runners, risers, sprues), followed by shot blasting and chipping of any mold flash. Then, the casting is inspected for any defects like blow holes, shrinkage defects, sand inclusions. On certain less critical castings, these could be lived with or repaired (Ni Rod welding). On a lathe bed, a homogenous casting free from defects is the only thing acceptable. With today's ultrasonic testing, it makes it possible to check a casting for internal flaws like shrinkage cracks that are not visible, or gas bubbles that might be opened up during machining. Years ago, castings got the visual inspection, and oftentimes, the defects were uncovered during machining. By then, it was too late, and a lot of shop time got wasted when the casting was scrapped.
Then, there is the matter of "seasoning" the castings. The old machine tool builders would take a run of castings for lathe beds, headstocks, carriages, tailstock bodies, etc and leave them outside in the weather. The castings stood out in the weather for a couple of years or more before they were machined. This was said to "season" the casting. The belief was the exposure to many freeze-thaw cycles and temperature swings, sun, dark, wind, snow, and all else in the way of weather would let the castings work out any internal stresses. Sometimes, the castings were stress relieved in a furnace, and then parked out in the yard for awhile. It was only after some form of stress relieving and/or seasoning that the castings were dragged into the shops and machining begun.
Some shops would rough machine a casting, then set it aside or do an in process stress relieving. After that, it was usually shot blasted and finish machined. machining a long casting like a lathe bed opened up a lot of "locked in stresses" in the casting, so it was predictable that the casting would "move" and what was a flat or true surface immediately after rough machining might not be there after a short time.
Making a good lathe bed (and the other cast parts) might take a few years if it were done the old way. Machine tool builders and automobile engine builders used to order the runs of castings a couple of years out from the time they would be machined and finished. Auto makers and machine tool builders would speak of "seasoned" castings. If you took delivery of, say, a 1935 model year car, despite the advertising as to the innovations in the 1935 model year engine, the block and head were probably cast in 1933 or 1934. The oldtimers had quite a belief about "waiting for the molecules to stop moving" in precision machine work.