When you are joining tables (after you split a table into two, you need to join them in a query), you can have a one to one, one to many or a many to many relationship. (The many to many requires a third table with 2 one to many joins - Access cannot do a many do many directly).
The most common is the one to many such as we would have in the example above.
A one to one join may be used for something like a table that has been split because it has too many fields. So it is split into a table with the most commonly used fields and another with the rarely used fields. Most of the time, you would use just the one table and select the fields you need. However, on a rare occasion you might need some of the info in the other table, so you would use both table's in a query.
An example of a many to many relationship would be a Reports table listing all reports in an Access database, a Criteria table listing all possible criteria for these reports, and a ReportCriteria table which contained ReportID, CriteriaID, LabelCaption, Top, Left, Width, DefaultValue, RequiredYN. This would be used to create a criteria form when a report is called. The OnOpen event of the report calls the criteria form and feeds it the name of the report. The OnOpen event of the form opens a recordset from a query that join all three tables and selects the records based on the name fed to it. It then loops through the records and places the textbox for each criteria at the specified place on the form and set the criteria's label to something readable instead of the name of the textbox, sets the width as specified (determined by possible values of that criteria for that report) and default value. Here you have a bunch of report that need several criteria and a bunch of criteria that are used by several reports. For example, almost all the reports required Start Date and End Date fields. These typically defaulted to the first day of this month one year ago and the last day of last month. One report required Prior Start Date and Prior End Date as well. These defaulted to one year prior to the Start Date and End Date values. About half the report required Office Number and most of these required Employee Number.
As you work with databases, you will develop the ability to see where thing need to be split. As far as bad ways to split a table. Hmm, let's see. How about a State table that looked like:
StateID, State
1,MN
2,VT
3,NH
4,MA
5,CT
6,RI
7,NY
etc.
That's taking normalization (splitting the tables is part of a process known as normalization) to an extreme. However, if I had the need to have the state name spelled out, I might want a table of state names, but I wouldn't use an autonumber for the key field; I'd use the abbreviation field as the key.
St, State
MN, Main
VT, Vermont
NH, New Hampshire
etc.
Then, of course, there are inner join and outer joins. An inner join says "Give me these fields for each record that has a matching record in both tables." If it exists in only one of the tables, you don't get anything for that record. An outer join says "Give me these fields from this table and, if there's a match in the other table, give me those fields also." In that case, you get all the records from the first table plus the fields from the other table if there's a match and a bunch of empty fields if there's no match.