The first communication a patent applicant may receive from the USPTO patent examiner is that the applicant must choose among different claims of the application in order to proceed with examination. This is a Restriction Requirement. Typically, a patent application is filed with 3 independent claims and a total of 20 independent and dependent claims. Each claim describes variations of the invention.
The basic rule is that each patent can claim only one distinct and independent invention. This is subject to the rules pertaining to restriction practice and the filing of divisional applications. (A related rule is that two patents can’t claim the same invention. This is subject of the rules prohibiting double patenting and stipulating to terminal disclaimers.)

First, the rule that each patent only claim a single invention appears contradictory to the encouraged practice that each patent contain multiple claims. As I have previously written, a patentable invention is defined by the claims. A claim can be either “independent”, i.e., it lists all the limitations or elements of the invention, or “dependent”, i.e., the claim incorporates by reference the limitations or elements listed in another claim but adds a further limitation or claim element to the description of the invention.

Each claim provides a definition of the invention. The multiple claims can be thought of providing alternate definitions of the same invention.
Adding to the complexity is that a single patent often includes claims for a method or process of making something as well as claims for the apparatus or object made by the process or utilizing the process. An example of a process claim and apparatus claim defining the same invention is if the process or method claims describe the invention as the novel method that may be used to create an apparatus or the apparatus claims describe the novel arrangement of components that comprise an inventive apparatus.

An independent invention is an invention that is not dependent or related to another invention. An independent invention is an invention that is unrelated in “design, operation or effect” to another subject disclosed in the specification (written description) or drawings of the application. This can be confusing because, in general terms, everything is related to something else.

The USPTO Manuel of Patent Examining Procedure (MPEP) tries to provide some explanation (See MPEP Chapter 8 at §806.04.) Two different combinations (of elements) are independent if they are not disclosed (in a patent application) as being capable of being used together, have different modes of operation, different functions or different effects. The two different combinations disclosed in a single application would therefore be subject of a restriction requirement. The applicant would be required to elect between the different combinations as to which combination would be searched and examined by the examiner for possible patentability. The non-elected combination could be separately pursed in a divisional application. (See below)

If it is determined that the inventions claimed within the patent application are not independent, i.e., related inventions, then the inventions must be analyzed for distinctiveness. (To muddle the matter further, the MPEP appears to use the term “patentably distinct” to describe both independent inventions and dependent inventions that are distinct.) Two inventions may be related but distinct if one is not obvious in relation to the other. Recall that novelty, utility and non-obviousness are the three essential elements of a patentable invention. (Obviousness is a separate topic for discussion.)

Basically, a patent examiner should not hold two related inventions distinct unless the examiner is prepared to approve two separate patents if the elements of novelty and utility can be demonstrated.
If a patent application discloses two (or more) patentably distinct inventions, i.e., two independent inventions or two related inventions that are not obvious variations of the other, then the patent examiner may issue a “restriction”. A restriction requires the applicant or inventor to “elect” one of the two or multiple inventions (again, as defined by the specific claims) for examination and prosecution. The non-elected claims may be either cancelled or merely withdrawn. The applicant is entitled to file a separate patent application, and pay a separate patent application fee, to prosecute the claims defining the non-elected invention. This second application is termed a “divisional application”. It is entitled to the same priority date as the original or parent application. Understand that at this stage, the parent application has been restricted to only contain claims to the elected invention.

The examiner is supposed to give reasons (not a mere conclusion) justifying the restriction requirement, i.e., the determination that the application discloses two or more patentably distinct inventions and forcing the applicant to elect among these inventions. This should allow the applicant to determine whether to object to the restriction requirement and preserve his/her rights. Whether or not the applicant objects (“traverses”) the restriction, the applicant is still required to make an election.

This brief article only begins the discussion of this topic. However, a restriction requirement may be the first action the applicant receives from the patent examiner. It will precede the examiner having made any search of the prior art. It may cause the applicant to make concessions regarding the invention(s) that may later be regretted (keeping in mind Festo and prosecution history estoppel).

Small comfort can be had in the USPTO conceding that the analysis or test of “patentably distinct” is “puzzling” and that only examiners having a requisite level of seniority and authority are permitted to issue restrictions.