In verterbrae:

Bone strength is maintained mainly by trabecular bone, while relative contribution of the cortical shell to whole bone strength remains poorly understood.  Cortical BMD reflects material density, but trabecular BMD is more significant predictor of vertebral fragility and fracture risk.

In femur, tibia:

Evalution of cortical BMD and trabecular BMD is of similar importance, as increased cortical porosity and decreased trabecular bone mass are most important causes of fragility.

view Chapter 2 (follet_Chap_02), page 26

First, designating bone as trabecular or cortical is an arbitrary distinction we make for convenience. In reality the body generates bone with a continuum of different densities, depending on location and function (amount and type of load bearing). So, there is no set level of density that differentiates trabecular and cortical bone. Somewhere in the literature, you may be able to find density ranges (hopefully not specific values) somebody has picked for trabecular and cortical bone, but again, they are arbitrary. 

I send you these files may be useful to answer your question.

see thesis of HELENE DUBAR: page 50

it is also worth noting that two different things can affect bone bulk (or apparent) density. One is of course the presence of voids, as in any bone specimen that would typically be described as trabecular. The other thing that can affect density is a defect in mineralization of the bone. In some diseases, the density of the bone per se (not the bulk density including voids) is below normal due to abnormally low mineralization. This is usually called osteopenia.

I am not sure, but there seems to be some chance that astronauts subjected to a long period of weightlessness may develop osteopenia as well as generally losing bone mass. Weightless astronauts lose as much as 2-3% bone mass per month. Even in normal gravity, forced immobilization affects bone mass. A person who must lie in bed for many days begins to have elevated calcium in the urine indicating a decline in bone mass. I have not researched how much of mass loss due to disuse atrophy is due to loss in bone cross section, increased osteoporosis or osteopenia. 

The question of the patho-physiology of the respective cortical and trabecular bone changes occurring during aging and disease is of course interesting but also, maybe inextricably, complex, as suggested by the above answers. The correctly cited difficulty to distinguish the two types of bone (and the possible futility of doing so) certainly compounds the problem. Moreover, any reduction in material density (as in osteomalacia), and any increase in cortical porosity, deterioration of trabecular micro-architecture and trabecularization of cortex  (changes occurring in endocrine diseases like hyperthyroidism and hyperparathyroidism that lead to accelerated bone turn-over) elicit adaptive macro-architectural changes in the cortex, namely periosteal bone apposition, and optimization of resistance to mechanical environment. Likewise, changes in mechanical environment, occurring  during prolonged bed rest and space flights, produce rapid bone loss and periosteal resorption; needless to say, in the opposite mechanical situation of mechanical loading like exercise, bone gain and periosteal apposition (modeling) occur. Bone tissue has plastic properties that adapt to endocrino-metabolic and mechanical perturbations so that bones can remain intact in the everyday life (even trying to maintain a conspicuous safety reserve).

One addition to the fine comments by Russo on effects of mechanical loading on bone amount and density:

Losses occur rapidly when little loading occurs (space flight, bed rest) and unfortunately regain of such bone loss occur slowly. For example, despite in-space exercise and high levels of exercise after space flight, losses incurred by astronauts during a few months in space are often not fully regained 3 years after a return to earth. This can be extended to indicate that a sedentary life style can produce bone losses that are extremely difficult to correct by exercise. 

I worked with osteoporosis (and all senile other signs) in my PhD thesis on paleopathology. The signs in examining directly specimens are: 1- they are first seen in spine (and also metaphyses of long bones, like femoral neck); 2- the cortical bone becomes much thinner, and with minute perforations, best seen with a stereoscopic microscope; 3- trabeculae are reduced in number, and show a very irregular pattern,  so that they become more distant each other. These signs are more evident when compared with a normal pattern. See my publications on paleopathology .    

From a clinical viewpoint, there is no need to differentiate between trabecular and cortical bone when measuring bone density. Typically, the vertebral bodies have more trabecular bone and femur more cortical bone. However, should either site become osteoporotic on measurement, the treatment would be broadly similar; there are some medications that are not so effective in preventing future hip fractures compared to vertebral fractures, so treatment needs to be individualised.

I would like to answer this crucial question by just making reference to the subject, with some but little regard to the consequences of possible misconceptions.

Quite different methodologies as quantitative computed tomography (either in its standard, "axial" version, QCT, in its "peripheral" adaptation (pQCT), or in its peripheral-high resolution modality, HR-pQCT), and ultrasound (US) have been applied to measure quite different kinds of "bone densities". My feeling is that I have to start by defining "density" both in physical terms and the way it is currently used by clinicians.

Classically, "density" is a measure of the concentration of mass per unit volume of a gas, liquid, or solid (d = m/v), and is expressed in the "cgs" system in g/cm3 units ("volumetric" density). This magnitude is useful to describe the differences of solutions o different quantities of the same substance in the same volume of a liquid, the different degree of porosity of a solid body, the different weights of the same volumes of different substances, etc.

By extension, this concept is also applied to the description of the concentration of anything on the surface of something, as a "mass (or any other suitable indicator or that, i.e. 'number of discrete elements', or so) per surface" unit ("areal" or "surface" density). In the cgs system this is commonly expressed in g/cm2 or number/cm2 units. This magnitude is useful to describe the "density" of hair on the skull skin, the "density" of inhabitants per km2 of land, the "density" of paint sprayed over a painted surface, etc. While this ("areal") magnitude has nothing to do with the former ("volumetric") concept, the use of the same word ("density") to refer to both of them has generated lots of confusion, especially among osteologists, with some catastrophic consequences concerning the concept, Pathophysiology, diagnosis, treatment and monitoring of osteoporosis.

I. Let's go first to (volumetric) "bone density".

   a. Bone matrix (volumetric) density. Grossly speaking, bone "matrix" is a mineralized mass of collagen. The (volumetric) density of the "pure" bone mineral, OH-apatite (3.2 g/cm3), is higher than that of collagen  (1.0 g/cm3), and the combination of the two in a normally mineralized matrix is close to 1.92 g/cm3. This is the only "true" (volumetric) bone density, and cannot be measured directly because it is impossible to get a measurable piece of matrix totally free of pores.

   b. Cortical bone tissue (volumetric) density. Cortical bone tissue is just bone mineralized matrix with a (volumetrically small) proportion of pores or lacunae, generally occupied by cells, canaliculi, and vessels. For that reason, its (volumetric) density is very close to that of the matrix. In practical terms, we can take it to be about 1.9 g/cm3, and directly refer to that as "solid" bone tissue density. As this (volumetric) density value takes into account both "true" bone (matrix) and also pores, it is referred to as a kind of "aparent" (Archimedean) bone density.

   c. Trabecular bone tissue (volumetric) density. In contrast wit cortical bone, trabecular bone tissue is a network of the above "solid" bone tissue (with its (volumetric) density of 1.9 g/cm3) spatially distributed by the bone mechanostat acording to biomechanical laws within a mass of bone marrow (with a (volumetric) density similar to that of water, i.e., 1.00). Thus, if we wish to talk about "trabecular bone (volumetric) density", we have to merasure the combined densities of the "solid" trabecular network and the marrow. Given the highly variable proportion in which trabecular network and marrow are naturally combined in different skeletal regions, we can not take a fixed reference to approach "trabecular (volumetric) bone density". Grossly speaking, we have just a wide range of values of that to refer to. This could rank, say, from somewhat less that the cortical density (1.9 g/cm3) to close to water density (1.0 g/cm3). As this (volumetric) density also takes into account both "solid" bone with pores and also marrow, it is also regarded as another kind of "apparent" (Archimedean) bone density.

   d. Whole (organ)-bone (volumetric) density. This is the (volumetric) density of the whole bulk of cortical and trabecular "solid" bone with their pores and marrow, vessels, etc that are contained in a bone integrated as an organ. Of course, there is a really wide range of variation of this (volumetric) density for different bones, so no "normal reference values" can be given for this further kind of "apparent" (Archimedean) bone density..

   Importantly, we must know that none of the above (volumetric) bone densities can be measured in clinical practice.

II. Now, we have to differentiate (volumetric) "bone density" as we have dealt it with above, from what we call (volumetric) "bone mineral density" (for short, vBMD). vBMD does not describe the "solid bone" mass contained in a given volume of bone tissue or in a whole bone, but just the "mineral" (mostly Ca, but also other elements) content ("bone mineral content", BMC) per unit volume of bone matrix, or cortical or trabecular tissue, or whole bones, expressed in g/cm3. In contrast with the "natural" bone densities referred to above, vBMD's are most useful variables in clinical osteology because they can be measured absorptiometrically (only by tomographic (QCT, pQCT, HR-pQCT), not by the standard "densitometric" (DEXA)  procedures) in many instances, with the obvious exception of the "matrix" ("true") vBMD (it is impossible to deprive the matrix of its tiny pores).

   a. Cortical vBMD. This is the BMC per unit volume of the cortical tissue, comprising both the matrix and its tiny pores. Being the volumetric proportion of pores so small, the vBMD can be regarded as the "true" vBMD of the "solid" bone. This is quite an important attribute of this variable, as long as the mineral (Ca) content of "solid" bone tissue has been shown to vary close to linearly with its intrinsic stiffness (elastic modulus) as a material concerning all compression, bending or torsion stress analysis in physiological conditions. Thus, we can (grossly) take the cortical vBMD as a biomechanical correlate of the "solid" bone tissue stiffness (i.e. as indicative of the "mechanical quality" of bone at the tissue (not organ) level. The normal values of cortical vBMD may vary from 1.1 to 1.2 mg/cm3.

   b. Trabecular vBMD. This is the BMC of trabecular tissue in the studied bone site per unit volume of all the "solid" bone network together with its tiny pores and the marrow. Being the volumetric proportion of marrow relatively high in this kind of bone tissue, trabecular vBMD must not be regarded as an indicator of the "solid" bone tissue quality as cortical vBMD is. Instead, the trabecular vBMD is a useful indicator of the degree of accumulation of "solid" bone tissue in the bone sites where it is present. Hence, while the cortical vBMD can be taken as a bone "quality" indicator, the trabecular vBMD is just a bone "mass" indicator. The normal values of trabecular vBMD may vary betweeen 0.25 to 0.35 g/cm3.

   c. Total (whole-bone) vBMD. This is the BMC of the sum of cortical and trabecular bone (if present) in the studied bone site per unit volume of all bone plus marrow space. Obviously, this variable varies a lot, depending on the proportions between both types of bone tissue and bone marrow in the studied site. Thus, there are no normal reference values for this variable, and its mechanical correlate is impossible to assess.

III. Let's go now to "areal" bone mineral density (aBMD). This is the BMC measured by standard (projection) densitometry DEXA per unit of projected surface. Obviously, this is not a "volumetric" but a "surface-related" density (as that of hair in the skull skin), and is naturally expressed per area unit, in g/cm2. It is measured classically in the lumbar spine, the femoral neck and trochanter and the wrist, and also in the whole body or in sections of it (skull, trunk, members). This is useful to assess "how much 'solid' (normal?) bone would be within a given bone region, with no discrimination between "cortical" and "trabecular" BMC or BMD", and with no volume specification. Widely spread reference charts of normal aBMD values for all the above regions, discriminated by age and gender, are available as acknowledged and recommended by the WHO. These values are given as Z-scores (regarding age ranges) and T-scores (with reference to normal samples of young, healthy individuals). Dhe DEXA T-scores were correctly proposed as indicators of the amount of bone mass in the studied regions and are most useful to propose a diagnosis of osteopenia (lack of bone mineral within the bones with normal "quality" of the remaining tissue), yet no information is provided concerning bone tissue "quality". By extension, the T-score range of DEXA-aBMD values has been wrongly applied to diagnose osteoporosis (osteopenic-derived bone fragility - NIH concept) despite that the aBMD has no mechanical correlate at all. Fortunately, this wrong WHO-approved assimilation is being progressively left aside in current Osteology.

IV. What about US?. On one side, BUA is just an approximation to bone mass measurement which usually shows moderate to good correlations with DEXA aBMD measurements in the same regions. It does not discriminate either between trabecular and cortical BMD and is just a correlate of DEXA, i.e. a very indirect indicator of a possible osteopenia. On the other side, SOS can provide some approach to the mechanical quality of cortical bone, whose mechanical cortrelate is still unclear.

V. Well, could now we go to the clinical meaningfulness of the described variables?

     a. Cortical vBMD estimates the intrinsic stiffness of "solid" bone tissue. Unfortunately, bone tomography is unable to differentiate between low cortical vBMD values induced by osteomalacia (deficit in bone tissue mineralization that canaffect bone tissue stiffness) and excess of intracurtical porosity (provoked by excessive intracortical remodeling that can affect bone tissue toughness). The same technology used to determine it (QCT in all its variants) can also measure indicators of bone tissue distribution in the studied bone cross-section, as the cross-sectional moments of inertia for bending or torsion (CSMI's). This allows 1. calculation of "Bone Strength Indices" (BSI's, SSI's) as functions of the product cortical vBMD * CSMI as indicators of the whole bone (structural) stiffness (see JL Ferretti's papers), and 2. assessment of the "distribution/quality" relationships between a given CSMI (y) and the cortical vBMD (x), as indicators of the functional status of bone mechanostat (see JL Ferretti's and HM Frost's papers). 

     b. Trabecular vBMD estimates the mass of trabecular bone in long-bones' epiphyses. This can provide a diagnosis of osteopenia specifically related to trabecular bone. The mechanical correlate of such osteopenia concerns the ability of the trabecular network to support the loads transmitted from the articular surfaces to the cortical walls of long bones, and to stand compression stress (within a much lower range of values than that supported by the surrounding cortical bone), i.e. to approach a diagnosis of osteoporosis, with a high dependence on the spacial alignment of the network with the deforming load.

    c. Trabecular network features can be described almost histomorphometricaly by HR-pQCT. This procedure can add interesting descriptive details to qualify any eventual trabecular osteopenia.

    d. "Projection" aBMD Z- and T-scores can diagnose osteopenia (not osteoporosis) better than any other described variable does, yet providing no distinction between cortical and trabecular bone tisues, and with no mechanical correlate at all.

    e. US has no mechanical correlate at all, and its BUA values would be always "secondarily related" to those of DEXA-assessed aBMD T-scores measured at the same sites.

    Sorry about that, I can add little more to the above to the diagnostic applications of any absoptiometric measurement of "trabecular" and/or "cortical" "areal" and "volumetric" densities.