linking tables for maximum efficiency

[buggirl]

Hi,

I am trying to build two tables containing information about fish that we
have collected:

tbl Fish:
FishPK (one for each individual fish; autonumber)
CollectionPK (links to tbl Collection which contains information about where
and when we collected the fish)
FishName
FishAge
FishSex
FishLength

tbl FishAnalysis:
FishPK
Analysis1
Analysis2
Analysis3
Analysis4

The chemical analyses are only performed on a subset of the entire fish
collection.

I am pretty happy with this design, but am open to suggestions! In
particular, I want to know if this design allows me to:

1. build a query to find the individual fish that have (or have not) been
used for chemical analyses

2. build a form for entering results of the analyses - when I receive the
data, it looks like this
Collection ID (not PK)
FishName
Analysis1
Analysis2
Analysis3
Analysis4
- which means that I have to collect data from two separate tables before
adding the new results to tbl FishAnalysis. How do I build a form to simplify
this process? And is my design correct to enable my form?

I would appreciate any input!

Thanks,

buggirl

 

[Jerry Whittle]

This part will cause you grief:

Analysis1
Analysis2
Analysis3
Analysis4

What happens when someone decides to add a 5th Analysis? All your forms,
reports, and queried based on this table will need to be modified.

You need this table to be different and add another table.

tbl FishAnalysis:
FishAnalysisPK (every table needs a PK even if just an autonumber)
FishPK FK to Fish
AnalysisPK (FK to the Analysis table!)
ChemoResult

tbl Analysis
AnalysisPK
AnalysisName
AnalysisNotes

This way you can add more chemical analysis in the future. Queries will be
much simpler to although you'll need to join in an extra table and remember
to do things like left or right joins to see all results.

BTW: An old buddy of mine, Jim Widlak, does work in a field like your
database suggests in Tennessee. A very tall guy and about 58 years old. Know
him?

 

[buggirl]

Thanks for your reply!

You know, I'm not too worried about adding a fifth analysis. I may not have
explained the data clearly enough. All of the analyses are conducted
simultaneously and so all of the results come together, as a set. If I did
another analysis, it would be conducted on different fish (as the current one
destroys the fish), so I would have to construct a separate table. (Otherwise
I would have a bunch of blank cells.) Does this make sense? And, if that's
the case, am I doing the right thing?!

Sorry, I don't know Widlack! I'm primarily a bug ecologist, and I'm just
learning about fish!

Thanks again!

 

[Jerry Whittle]

One good rule to follow about database design is to build for easy expansion.
You may only have 4 analysis now and maybe forever, but there's heck to pay
the day someone adds a 5th one. I've learned and relearned this hard lesson
more than once especially after a customer swore that it would never change.

Now it is possible to go overboard. Take phone numbers for example. Back in
the day an employee had A phone number. Then someone added Fax numbers. Then
the Home phone numbers. Then pager numbers. Now Cell phone numbers. You could
make a case that all these phone numbers should be in a seperate table
instead of all in different fields in the Employee table. In fact I have a
database set up like that. However it's often easier to de-normalize the
database just a little in cases such as phone numbers.

It's up to you.

 

[Armen Stein]

One good rule to follow about database design is to build for easy expansion.
You may only have 4 analysis now and maybe forever, but there's heck to pay
the day someone adds a 5th one. I've learned and relearned this hard lesson
more than once especially after a customer swore that it would never change.

Yes, and even if you're not worried about future expansion (I've been
in this business too long not to worry about it then you still
should consider that querying 4 different fields in 1 record is always
harder than querying 1 field in 4 records. Databases lend themselves
to sets of records, not sets of fields.

Armen Stein
Microsoft Access MVP
www.JStreetTech.com