When more than one transformer is used on a toy train layout, it should be phased with the other transformer. This is easy with the post-war Lionel ZW transformers, as their U-posts are interconnected and are clearly the common for all their four outputs. [Speaking technically, all the four outputs have the same phase angle relative to this common U-post, which is why the procedure is called "phasing". That procedure, with a 2d postwar ZW, is to insert its house-voltage line plug in such a way that the second set of outputs will have the same phase angle as the first. The two sets of U-posts are linked by a tie wire or the common rail, and only two outcomes are possible, either in-phase, or 180-degrees out of phase.]
Being in phase will prevent any voltage on the layout source posts from exceeding 30-volts in their difference. In any locale with 120v nominal house voltage (permitted to fluctuate to 5% overvoltage), the postwar ZW will with whistle activated produce at most 29.91v*. This is the so-called root-mean-square voltage of a sine wave whose peak is thus 42.42v. The danger is known to lie in the peak voltage, and the danger zone had long been considered to start at 44 volts peak.** This is a matter of life-safety. You will want always to observe your life-safety considerations, despite any convenience in doing otherwise. 
If the second transformer is dissimilar to a postwar ZW, one might have to ask the help of a knowledgeable train repairman to learn the appropriate common post. For the 90w 1033, the appropriate common is the A-post, as here the U-post indicates a variable output [possibly in continuation of a practice from the early days where a fixed output was indicated XYZ for 3 phase power, and TUV for a variable output (again 3 phase, the letters W and S being possibly used for a 4th wire]; this results in all other 1033 outputs having the same phase angle relative to the A post, and is contrary to the published instruction to connect the 1033 U-post to the outside rail (relative to the widespread ZW practice of connecting ZW-U to an outside rail at present). I believe the overriding reason for making the ZW, despite its output amps making it ineligible for a UL-label, was to avoid the need for a second transformer and the possibility of touch-voltage accidents, for many, possibly most layouts at the time.
The MTH Z4000 produces its track outputs from a 28v rms sine wave, I have been told. Thus in a 120v area, the maximum expected output would be 29.40v; the corresponding peak is gotten by multiplying this by 1.414. Footnotes follow:
* The whistle coil produced 6v in the115v house voltage common at the time of its design. BTW, the postwar ZW had a bit of added iron in the core so that it could be operated at the 125v Mexican voltage, which was also then used in parts of southern California (125/115 as much iron for that, plus the addition for 5% overvoltage, plus probably an unknown amount (10 or 15%***) to prevent burnouts from this cause (ie, excessive overvoltage).
** The problem at 44 volts was known from pioneering work with DC distribution. Today, there is more research that indicates the effects of AC and DC differ somewhat; I do not know of any work with AC having a DC component, as is the case here with the whistle signal. The NEC (national code) used both 30 and 32 volts rms as limits in various sections; very recently all these limits were made the same at 30 volts (AC sine wave).
*** An AC induction motor operates on the transformer effect, and is provided with extra iron to survive unusual overvoltages, 10% being fairly common, 15% being a better quality; a quality code is often stamped on the nameplate. Unusual undervoltage is also a problem with motors, requiring more copper (or aluminum).
Frank
