Historically, taxonomy has been the playground of débutants and the workshop of balls-to-the-wall genetic systematics. It is the method by which names are applied to organisms, to distinguish concepts of how life can be categorized. We’ve seen different systems of nomenclature (the actual subject of names) applied to phylogeny (which together form taxonomy), and we’ve touched on the various ways in which taxa have been regarded as dubious, set aside, or expanded on the basis of the concept of the nomen dubium.
Problematically, there is no “one way” when it comes to regarding something as a nomen dubium, and the practice of regarding something as one, as how you treat it afterwards, has been variable historically. I’ve had plenty of opportunity to regard the various concepts, and their applications, and realized a few things about the phrase and label (and how its been differentially applied to various taxonomies).
First, systematists working on bulk taxonomic lists, without special attention to some, will use the phrase to apply to anything that they cannot place to some particular level. This is especially important in considering Linnaean systematics, in which a “Genus” will be regarded as a nomen dubium because it cannot be placed into a “Family.” The method has been followed despite the loss of use of the term “Rank,” or even which ranks are being used, but followed nonetheless. For the most part, this is because of the subjective inference given to the identity of the “Genus” above that of the “Species,” and the need to validate this higher up; so that while “Family” is not mentioned it is implied.
Second, systematists attempting to describe a fauna (a specific set of animals in a specific time) will often be faced with material that has been regarded by previous workers, and some of those workers would have named material. If that material is spotty or incomplete, and later material shows up that is far more complete, such a systematist will hedge his bets and attempt to set aside previous taxonomy in order to preserve the idea of the more complete material as the basis of new taxonomy. This is more correct in a sense that we should endeavor to more complete type specimens to base our taxonomy on, but it ignores some details of the particular (how complete does material have to be to be diagnostic? or how many features does something have to exhibit to be diagnostic?) that are unquantified.
Third, systematists will set aside taxonomy when they regard the number of features as being “too little” in the way of diagnostic value in the diagnostic section of their analysis (or previous analyses, if they regard them at all), and label the resultant “shelved” taxa as nomina dubia.
So what is a nomen dubium?
The Code for the ICZN (the International Commission on Zoological Nomenclature) uses the following definition:
A Latin term meaning “a name of unknown or doubtful application”.
Historically, the ICZN was a little more clear, as were the systematists who originally used it, for which  provides a historic discussion. But this does not help us as much as we like. It does not tell us how to apply it, unlike the sections in the Code on types. Instead, we are given an ambiguous label for an ambiguous concept, and it is applied inconsistently, often seeming to be in keeping with an intention to erect or set aside nomenclature, and sometimes both.
Much of the discussion, unfortunately refers to a the larger discussion on the term “diagnostic” and how that applies to taxonomy. I won’t get into that here. Importantly, it can be regarded that two hard and fast rules for the treatment of nomina dubia (as  details) is
1. Taxa indicated as nomina dubia cannot be used as the basis of further taxonomic acts;
2. Following this, taxa indicated as nomina dubia which are the basis of taxonomic acts should have those contained or containing taxonomies replaced.
While 1 is often followed, 2 is seldom, and this is because it becomes more confusing to use nomenclature when whole systems can be set aside. The ICZN allows for such destabilization (up to the “Family” level) by permitting the retypification of nomenclature, thus preserving content and history, while removing the dubious material. This generally involves the use of neotypes, and designation of new type species when the original cannot be maintained.
Were we to apply the process of 2 more often, nomenclature would shift significantly whenever a new worker decided that certain fauna or taxa were in need of a new treatment. This is apparent in most treatments of continental, country, or formational investigations, and often highlights the previous authors’ tendencies to either name poorly preserved or incomplete material, or the provisional referral of material to other nomenclature to prevent them from naming such taxa (and put the burden of new, “dubious” taxa onto someone else).
What is “Diagnostic”?
Ultimately, a nomen dubium label rests on the author’s regard for the diagnostic value of the material on which the taxon is based. If this continues despite the referral of further material, then it becomes more difficult to establish the taxonomy. Because of this, the term “diagnostic” requires treatment:
“Diagnostic,” unlike other scientific terms, differs in use depending on the field, and its use in paleontology (as in biological description). For the most part, “diagnosis” follows differential diagnosis: The composition of features is compared to the composition of features in other subjects; the more the particular subject differs, the more unique the subject it; conversely, the less the subject differs, the more likely that subject is to be to the most similar differential subject. This is a more formulaic treatment, but it’s both applicable to medicine as it is to zoological comparison.
In the specific, “diagnostic” typically refers to a feature that is discernible. When considering taxonomy, the ICZN offers this restatement of the differential diagnosis (above):
A statement in words that purports to give those characters which differentiate the taxon from other taxa with which it is likely to be confused.
Here, it is a list of characters that can differentiate the taxon from something else. When dealing with such a definition, there are three levels of differentiation:
1. Singular differentiation: The specimen has one or more features which differentiate it from one possible relative.
2. Compound differentiation: The specimen has one or more features which differentiate it from two or more possible relatives. (In this and the above, many such features generally involve proportions, the shape of a process, the position of a process, etc.)
3. Class differentiation: The specimen has one or more features which differentiate it from a broad, almost universal list of possible relatives. (Many such features may be neomorphic bones, particularly unique ornamentation, unique cranial nerve relationships, etc.)
Diagnostic features are also classed based on their prevalence among other taxa, and are called apomorphies. A typical apomorphy is simply a feature, any feature, and any quality of a bone, or bone-to-bone relationship, even the ossification of a non-bone element (e.g., a tendon). Synapomorphies are useless for differential diagnosis, as they refer to features shared between taxa. Autapomorphies are features that, while they may be present in other taxa, are unique to the subject compared to closely related taxa, and they can.
While some diagnoses regard only the presence of autapomorphies, others utilize a more collective approach, and use lists of apomorphies and different taxa in order to create the general differential diagnosis. Sometimes a combined approach is used. What is more interesting, though, is what is considered sufficiently diagnostic to produce nomenclature; and on that, no one has agreed upon a specific, limiting definition.
More problematic is that the term has also been used regardless of the number of features (apomorphies) that can be discerned, based on the seeming completeness of the material. It is almost certainly logical to assume that the amount of material is proportionate to the number of apomorphies that can be discerned, but it is illogical to assume that one must have a standard volume of material to have features. Some features may only be categorically verified on the basis of several complete skeletons (e.g., Archaeopteryx lithographica) but other taxa are based on very, very small portions of the total body (especially conodonts, which are often diagnosed on one or two pieces of a 10+ element inter-oral apparatus). Some taxa may be more diagnostically viable on the basis of certain regions of the skeleton, lacking statistically “significant” variation in others (e.g., centrosaurine ceratopsids, i.e., the “horned dinosaurs,” which have been called skeletally conservative, but cranially variable). This makes the argument from logical being applicable to all taxa problematic, especially if any of the above taxonomic groups are truly limited in their diagnostic potential. Some taxa, therefore, are likely to be more variable across their skeletons, and evolutionary processes typically are not limited to singular portions of the skeleton.
When it comes to being “diagnostic” enough, some authors have indicated various solutions. A specimen is diagnostic enough if it is
1. capable of producing an autapomorphy;
2. capable of producing any apomorphy, rather than just an apomorphy;
3. producing features that differentiate the taxon substantially from at least TWO different taxa;
4. having a complex of features, regardless of their differential composition, so far as the series of features can be differentiable as a unit (an “apomorphy suite”);
5. seems special.
Historic taxonomy involves most of these, but as the quality (and quantity) of material increases in time, perspectives on diagnosticity (how diagnostic is it?) have changed, favoring more and more complete material; regarding the potential of less complete material, even uniquely shaped structures, as less useful. This is, in part, due to individual variation or, in the case of teeth, variation along the row (or rows) of shape, size, and features of teeth. The balance between individual variation and diagnostic expression requires substantial material, and few fossil taxa provide the means to test this.
 Mones, A. 1989. Nomen dubium vs. nomen vanum. Journal of Vertebrate Paleontology 9(2):232-234.