diff --git a/docs/developers/adr/002_structure_for_describing_macroscopic_structure.md b/docs/developers/adr/002_structure_for_describing_macroscopic_structure.md
new file mode 100644
index 0000000..b10e7ef
--- /dev/null
+++ b/docs/developers/adr/002_structure_for_describing_macroscopic_structure.md
@@ -0,0 +1,24 @@
+# A Structure for Describing Macroscopic Structure
+
+## Context
+
+The main job of spinW is to calculate the energy levels of a magnetic system over a range of q values. This
+task is relatively self-contained and well defined by just the material's cystallographic and magnetic properties.
+
+However, there are a number of things that are more properly understood as a property of a particular experiment. 
+The main features are to include here is whether or not the crystal is twinned or is a powder.
+
+There is a question about whether it should be the location for the option to apply neutron scattering selection rules.
+
+
+## Decision
+
+We will create a base `Sample` class with derived `TwinnedSample` and `PowderSample` classes which will have methods for preprocessing the input **Q**-vector list and post-processing the output spectra from `spinwave()` for the case of twinned crystals and powder (polycrystalline) samples. The base class will be used by default for untwinned crystals.
+
+## Status
+
+Accepted
+
+## Consequences
+
+This will add another new object that users are unfamiliar with. (see backwards compatability proposal) 
\ No newline at end of file