I cannot find a convention 6 in this paper

I do not see why we enforce this.

It is quite unclear to me what is supposed to be denoted here. The unit vector along the the first axis that spans the coordinate system has the coordinates [1,0,0], right (?) Same for 'y', 'z'.

That is saying we do not give this information in NXcollectioncolumn and NXenergydispersion anymore? (Ok with me.) Also, we do not require information on the orientation of the hemisphere, as this is too special | only relevant for hemispherical analysers?

About geometry: In my datasets (and I think also in Stephan's) we have a group 'base_vectors' that contains the directions of gravity and photon beam in the lab frame. I do think this is needed to understand the subsequent NXtransformations-groups.

Strictly speaking, for the way we use this field 'electric_field' is a misnomer, since we mean the electric potential. I assume this was always the intention (of NeXus authors). NXmpes now introduces the new field 'bias', which is what we actually mean. I suggest to omit the 'electric_field'-item. Ideally, we should clarify on NIAC-level how these fields are supposed to be used in the future.

?

... or add a corresponding solute-group (see below).

Vorschlag: For flat jets, this group may appear more than once. If a mixture of solvents is used, the one with the highest mol fraction should be designated solvent, and the other(s) solute (see below).

cylindrical

point of injection

through

Cut-and-paste error

Clarify that this is background gas pressure while the experiment was done. (An interesting situation arises for gas-phase PES! Have no good suggestion for this, yet.)

As said earlier, I do not think it is useful to make this a mandatory field and would strongly dislike to implement it in our data records. For the EU realm, substantial documentation effort would be needed to make these entries compliant with DSGVO, or they would have to be stripped when posting data sets on public repositories, such as zenodo. I again suggest to cut it or make it optional.

a) What is the intended use of this item, and in which way does it differ from the NXdata-group raw? b) I do not think that it is useful to tie down the dimensionality of this field, as this in practice will depend on capabilities of the set-up.

I wonder if what is written here is actually meant: We have an NXdata-group with arbitrary name, which contains a data item, say an N-dimensional array, of measured values called 'raw'. I would prefer to have the group called 'raw' and the data item something more descriptive, say 'intensity'. Same goes for the NXdata group below.

It should be allowed or even required to have an NXtransformations group to indicate the direction of the analyzer lens. (Currently, it could be introduced via an NXlens group, which could have an NXtransformations field, but I think this is overly complicated.) (For a hemispherical analyser, another vector is needed to fully describe the orientation of the hemispere (fix rotation around the lens axis).)

Meaning of this field is unclear to me: My understanding is that it will be used to record residual gas pressure in which the sample is placed. It seems however that the field already exists in standard NeXus, but with the meaning of an external pressure applied to a sample while the experiment is conducted. If all of this correct, rename to 'residual_pressure' ?

Field 'chemical_formula' is required, but our samples are mixtures of different chemicals (solute & solvent), so a single chemical formula is not useful to give. We list chemical formulae of ingredients in a set of NXsample_component groups. This should be allowed as an alternative to giving a single formula.

In my routines, the fields fast_axes and slow_axes are currently auto-converted into str-arrays as they are written like arrays. Of course, with additional effort this could be prevented - I wonder however if the intent was really to have here something like "[energy, detector_y]", including the square brackets, as a single string?