Our IT person confirmed that our firmware is up-to-date and that OCR is disabled. I manually tried high compression and found that it made very little difference (the warning displayed to me also said that it has minimal impact on non-photographic content).
I am currenly sending the scans to anoher network, extracting the images, reducing the grayscale levels, and recombining them into a much, much smaller PDF. Looking ahead to the aimed-for paperless office, I find that the scan capability's value is severely compromised due to the onerousness of this requirement.
As I have described, the scanner's original PDF is much bigger than the sum of the sizes of the image files contained therein; decreasing the number of grayscale levels aside, therefore, it looks like a large contributor to the problem is the manner in which the original PDF is assembled by the scanner from scanned images.